Hypotheses

Page: https://fernhill.example/checkout. Primary metric: completed purchase rate of checkout page visitors.

Each hypothesis names one change and a direction to test. None promises a lift.

Sample size and duration

Inputs: baseline p1 = 3.2% (0.032), MDE = 10% relative, two sided alpha 0.05 (z_a = 1.96), power 0.80 (z_b = 0.8416), 2 arms, 18,000 weekly checkout visitors. Computed by script.

• p2 = 0.032 x (1 + 0.10) = 0.0352

Good material to work with: confusion about the annual toggle and an unclear enterprise tier are both clear things to test on the pricing page. To write testable hypotheses with real sample sizes, I need two required numbers:

Flow overview

Trigger: checkout started in Shopify, no order placed. Platform: Klaviyo. Timing is taken from your input.

Email drafts

Limits: subject under 45 characters, body under 120 words. Counts were checked by script; body counts exclude dynamic blocks and the CTA label. Voice: calm and plain, with no exclamation marks.

Email 1 (4 hours)

• Subject: The items in your Alder & Loom cart (35 characters)
• Preview text: Return to checkout when you're ready.
• Body (20 words):

Hello,

I can write this flow for Alder & Loom. Three items need confirming first, because cart emails quote your policies exactly and a wrong number would be in front of every customer:

1. Free shipping threshold. You wrote "over $50, or is it $75 now". Which one is live on your site today? I won't pick one for you.
2. Returns policy as published. "Returns are easy" isn't something I can quote. What's the return window, are there conditions (for example unused items only), and who pays for return shipping?

What I read

35 weekly rows for 5 ads over W1 to W7, plus each ad set's creative count and current Delivery status. A result is a purchase. Baseline is W1 and W2 pooled, as you set. You said there were no budget, audience or offer changes in these weeks, so I did not look for those causes.

Limits first: this is 7 weeks per ad and one row per ad per week, so the trends below rest on 7 points each and are a reading of your data, not proof. Meta marks frequency as estimated ("This metric is estimated").

Rising cost with a frequency near 3 could be fatigue, or a budget, audience or offer change. I can't tell from one frequency figure, and Meta's pages give no frequency number where fatigue starts, so 3 on its own does not answer it. To compare each ad with its own early weeks, I need three things:

Block table

'Search - Emergency Plumbing', Halcyon Plumbing, 2026-08-01 to 2026-10-28, bid strategy Manual CPC. Rules: CVR = conversions / clicks; CPA = spend / conversions; totals from summed rows; CPA change = block CPA minus total CPA. Computed by script.

Data sufficiency

I can plan a schedule from your data, but I cannot set anything up before I see it, and your bid strategy changes what I can propose.

1. Day and hour data (required). Per day and hour block for the campaign: spend, clicks and conversions, for one date range. Why: I compute conversion rate and CPA per block and only act on blocks with enough clicks.

Weekly portfolio rollup (internal, agency team only)

Inputs used: the fictional test data you pasted. Three weeks per client (two weeks ago, last week, this week), targets for all four clients, weekly spend plan, and expected channels. No account owners were given, so action prompts are addressed to "Account owner".

Happy to build the portfolio view. Before I start I need two things, because each client is scored red, amber or green against its target using real numbers, and I will not guess either:

Summit Physical Therapy: October monthly report

Prepared from the fictional test data supplied (October vs September). All numbers come from that data; derived figures (totals, changes, cost per booking) are calculated and shown with their math.

Glad to write it. Before I start I need three things, because the report leads with results against the client's goals and I only use real numbers:

1. Channel data for this month and last month: per channel (for example Google Ads, Meta Ads, organic, email), spend, results such as leads, bookings or sales, and revenue if tracked. CSV exports or pasted figures are fine.

AI answer visibility audit: Ledgerline

Limits first: these are 20 answers you collected on one day. Assistant answers vary by session, account, location and day, so results move. Every figure below is a count from this sample, not market share. I did not run either assistant; I scored only the text you pasted and wrote no answers. All answers are marked fictional test data in your input.

1. Method: date, assistants, question count

I can measure that, but I only report what the assistants actually said, so I need a few things first:

1. Your brand: name, website domain, and a one line description (I use it to spot wrong descriptions).
2. Competitors: two to five competitor names, so I can see who shows up where you do not.

Groups found

Other line kept as is: Sitemap: https://northwind-recipes.example/sitemap.xml

Token check

Google's robots.txt guide says a user agent "can match only one rule set" (the first, most specific group that matches it). So a crawler with its own group does not also follow the * group.

Conflicts

I can review this, but I need your goals before I write any new file, and I will not push anything live.

Answer sentences found

Target question: how long does a heat pump last? Queries you gave: heat pump lifespan, how long do heat pumps last.

The page answers from Ridgeway HVAC's own records only. It gives no general figure for all heat pumps, and I do not add one.

New opening

I can rewrite your pricing page so its answer comes first, but I need you to settle four items before I start: two required inputs, one claim I cannot write without a source, and one optional input.

1. Page text (required). Please paste the current pricing page, headings included. I only move and tighten facts already on the page; I do not write new ones.
2. Target question (required). Which question should the opening answer, in your words? For example "how much does [product] cost per month?"

Domains

Campaign 'Search - Running Shoes' (stridelab.example). Period A 2026-09-01 to 09-30 (30 days), Period B 2026-10-01 to 10-28 (28 days). The periods are not equal length; you asked to compare anyway, so every change below compares rates, not totals, and should be read with that gap in mind.

Change table

Change = B minus A, in percentage points, computed by script.

Entered and left

I can compare auction insights periods, but I cannot tell you how much a competitor spends. The report does not show that, and I will not guess it.

To find what changed, I need:

What I used

32 campaigns, last 30 days, from your paste. Goal: "demo signups at or under $150 each". Tracking: Google tag conversion tracking, Primary action Demo signup. Conversion value is 0 everywhere, so ROAS and Target ROAS are out of scope. Review rule: the default, labeled: Google recommends measuring Target CPA performance over "the last 30 days, including at least 30 conversions".

Per campaign table

CPA = cost / conversions, by script, blank at 0 conversions. Compared with your $150 goal.

I can help pick one per campaign, but the answer depends on each campaign's data and your goal, so I need three things:

1. Per campaign, the last 30 days: name, current bid strategy, conversions, cost, and conversion value if you track it. For all 5 campaigns, so I can compute cost per conversion and return on ad spend for each.

X thread draft: "Why we stopped discounting"

Source: pasted article text, https://goodgrain.test/blog/no-discounts (fictional test data). Every number is copied exactly from the article.

Defaults used: no account_voice was given, so the voice is plain first person plural ("we"), matching the article. No link_placement was given, so the link goes in the last post (default). No emojis.

1. Claim and points extracted

Happy to turn it into a thread of 6 to 12 posts (fewer only if the post is short; I never pad with outside facts). I cannot see which post is your latest from here, so I need:

1. The article: paste the full text, or send the URL. If you send a URL and I cannot open it, I will ask you to paste the text, because every fact in the thread has to come from the post itself.

Optional:

Brand entity consistency check: Northbeam Tutors

Checked: 2026-10-29, from the fields you pasted (fictional test data). I did not open the live pages, so every result reflects the pasted text as of today.

Canonical facts (from you): name Northbeam Tutors; legal name Northbeam Learning LLC; description "online math tutoring for grades 6 to 12"; founded 2021; HQ Columbus OH; founder Ana Ruiz; website https://northbeam.test; logo https://northbeam.test/logo.png.

I can check that. I will not infer your official facts from your profiles, because a profile might be the one that is wrong, so first I need:

1. Canonical facts (your approved version). Required: the exact brand name (is it "Northbeam", or something longer?), a one line description, and your website URL. Also send, if you have them: legal name, founding year, HQ city and state, founders, and logo URL.

Dead URL table

Site: https://www.copperkettle.example. Data: Ahrefs Best by links, 404 filter, 2026-10-28. min_ref_domains: 2 (from your input).

Rows: 6. Redirect: 4 (D1 to D4). Leave as 404: 2 (D5, D6). Check: 4 + 2 = 6.

Redirect map

Each rule is a server side permanent redirect (301 or 308). Google's redirect guidance treats server side permanent redirects as the clearest signal, and its site move guidance says to keep redirects in place for at least 1 year.

I can map your migration 404s to live pages, but I will not send them all to the home page, and I need some inputs before I start.

Why not the home page: each dead URL should go to the live page on the same topic. If there is no such page it stays a 404. Sending unrelated URLs to the home page drops visitors somewhere that does not match the link they clicked.

Required:

Snapshot

Shiftly vs RosterBee, for account executives (default audience). Q3 2026: 18 deals, won 11, lost 7; win rate = 11 / 18 = 61.1%. Oldest fact 2026-10-20; recheck by 2027-01-18 (oldest + 90 days).

Where we win

1. Toast. Shiftly integrates with Toast; RosterBee lists Square and Clover only. Top won reason: 6 of 11 wins.
2. Tip pooling. Included at $3; RosterBee puts it on Pro at $4, so $4 - $3 = $1 less per employee per month. 4 of 11 wins.

I can build the RosterBee card, but each competitor claim on a battlecard needs a source and a date. Reps lose deals, and trust, when a card says something a prospect can disprove. Here's what I need:

1. Sourced RosterBee facts (required). Right now all three claims are unverified, so I'd leave them off the card:
2. "Cheaper": cheaper by how much, on which plan? A pricing page URL and the date you checked it is enough.

Verdict

Needs fixes. Counted from the checklist table below: 8 pass, 1 partial, 1 fail. Rule: ready if 0 fails and at most 2 partials; needs fixes if 1 to 3 fails; return to strategist if 4 or more fails. With 1 fail, this brief needs fixes before it goes to a new freelancer.

Checklist results

Default 10 point checklist. Writer level: new freelancer.

Questions for the strategist

Before I score this, two things need sorting out, because both change the result:

1. The brief looks cut off. The Sources line ends mid sentence: "see the doc from last quarter and the". Please paste the rest of the brief. Also send the doc itself or a link to it, since "the doc from last quarter" isn't something a new writer can find. If more lines follow (goal, word count, due date, approver), I need those too. Scoring a partial brief would mark items as missing that might be there.

Content decay refresh plan

1. Periods compared

• Current period: Jul to Sep 2026. Previous period: Apr to Jun 2026. Both are 3 months, so they are equal length.
• This is a consecutive comparison, not year over year, so seasonal swings between spring and summer can look like decay.
• Data: your page export, page details and top query per page (fictional test data). min_clicks default 50. No conversion data, so ranking is by recoverable clicks only.

I can rank them, but I need the data first, because I only diagnose decay from real Search Console numbers:

1. Current period export: a Search Console Pages export (page, clicks, impressions, CTR, position) for the last 3 months.
2. Comparison period export: the same export for an equal length earlier period. You said traffic is down "this year", so the same 3 months a year earlier is the best match; the prior 3 months also works but can mix in seasonal swings. Which comparison do you want?

Status split

Statuses kept exactly as Search Console shows them. URL Inspection live test: not run (from your input).

Signals per URL

Median word count of the 6 pasted URLs = 855 (median of 298, 305, 310, 1,400, 1,820, 1,850, computed by script). Every signal is a lead to check, not a proven cause. My sources give no word count threshold for indexing; the median is only a way to compare these pages with each other.

Groups

URLs with the same set of signals.

I can sort your not indexed pages by likely cause, but I need the data first, and I cannot get pages indexed this week or on any date. Google decides that. Here is what I need:

1. URL list (required). From Search Console, Page indexing report: the URLs and the exact status shown for each (for example "Crawled - currently not indexed" or "Discovered - currently not indexed"). "Half our site" could be several statuses with different meanings, so I need the exact labels.

Case study draft: Harbor Freight Movers and RouteJet

Status: draft for internal review. The customer approved being named (per your input). Every quote and number must still be approved by the customer before publishing. Source: interview notes and confirmed metrics you provided (fictional test data).

Source key (the notes have no line numbers, so lines are numbered in the order given): - N1: Dana (ops lead): "Before, I built every schedule in a spreadsheet at 6am."

Happy to draft it. To keep every claim real and approvable, I need:

1. The interview: a transcript or your notes from a conversation with someone at Acme, with who said what (name and role). I only use quotes that appear in it.
2. Customer details: Acme's industry and size, and has Acme approved being named? If that is not confirmed, I will write the draft anonymized and flag it.
3. Product: which of your products or plans Acme used.

Optional:

Contrast table

Juniper Yarn Co, 'October restock'. Ratios computed by script with the WCAG relative luminance formula (sRGB channels linearized, L = 0.2126 R + 0.7152 G + 0.0722 B, ratio = (lighter + 0.05) / (darker + 0.05)); that implementation is mine. px to pt uses 1pt = 1.333px (my conversion). WCAG 1.4.3 asks for a "contrast ratio of at least 4.5:1", or 3:1 for large text, which it defines as "18 point text or 14 point bold".

Alt text

Checked against the W3C alt decision tree.

I can audit the newsletter for accessibility and dark mode risks, but I need the template details first. I will not call it "accessible" or "dark mode proof" at the end; I will tell you which checks passed and which to test.

Deliverability health check: brightmail.test

Source: the records and stats you pasted (fictional test data). No live DNS lookup was run; every result reflects the pasted records as of 2026-10-29. MX records were not provided. ESP: Mailchimp. DKIM selector: k2, as shown in your Mailchimp domain settings.

1. Records found (quoted)

2. Pass, warn or fail per check

3. Signals review

Last 30 days, from your stats:

Let's find out. Spam placement usually comes down to authentication records and sending signals, and I only judge records I can actually see, so I need:

1. Sending domain: the domain in your From address (for example yourbrand.com).
2. Email platform (ESP): which tool sends the emails, for example Mailchimp, Klaviyo, HubSpot, or Google Workspace?

Optional, but they make the check much more useful:

Send table

Fieldnote Coffee, Klaviyo, weekly Tuesday newsletter. Rules: open rate = unique opens / delivered; open rate excluding MPP = (unique opens minus MPP opens) / delivered; click rate = unique clicks / delivered; bounce rate = bounces / (delivered + bounces); complaint rate = spam complaints / delivered; unsubscribe rate = unsubscribes / delivered. Segment labels are as given.

Before and after

Period rates use summed counts over summed delivered, not averages of the send rates.

Hypotheses ranked

I can work out what moved, but I need the raw counts and the details of the domain change first. Right now there are two possible causes, and the data you sent can't tell them apart.

What I need (required):

Source table

Harbor & Pine Candles, Klaviyo, 2026-09-01 to 2026-09-30. Labels exactly as exported. Rules: net = added minus removed; removal share = removed / added; early unsubscribe share = unsubscribes within 30 days of joining / added.

Totals: added 820 + 140 + 410 + 300 + 120 = 1,790; removed 64 + 6 + 22 + 96 + 18 = 206; net 1,790 - 206 = 1,584. All source removal share 206 / 1,790 = 11.5%.

List level check. Rule: gap = source net minus list size change.

Happy to audit your Mailchimp list growth. Before I start, four things are missing or unclear, and each changes the numbers.

1. The report period (required). "Last quarter" could be July to September or a fiscal quarter. Please give the exact start and end dates, so adds and removals line up with the form numbers.

Email Pre-send QA Check: Larkspur Pantry autumn blends email

Platform: Klaviyo. Gmail volume you gave: about 7,500 messages a day. Tracking rule: links to larkspurpantry.example carry utm_source=klaviyo and utm_campaign=autumn-blends.

1. Link check

Checked by script per link: scheme, path against the intended destination you wrote, tracking parameters, duplicates.

• L3: add utm_campaign=autumn-blends.
• L4: the link starts with http, not https. Change it to https.

I can QA it within the hour, but "make sure it's fine" needs the email itself and what each link should do. I need three things:

1. The email exactly as written: subject, preview text and body, with personalization tags as they appear (for example a first name tag). I check each tag for a fallback so no line reads with a blank.
2. Every link: its label, the URL as written, and where it should go. Without the intended destination I can only mark a link 'intent unknown'.

RPR table

Pinecrest Coffee, 2026-09-01 to 2026-09-30, conversion metric Placed Order, email only. Rule: RPR = attributed revenue / delivered; Klaviyo says it divides revenue "by the number of successful deliveries". Computed by script at full precision, shown to 2 decimals; change = row RPR minus its type RPR, from the shown values.

Totals by type

Rule: type RPR = sum of revenue / sum of delivered for that type, not an average of row RPRs.

I can compute revenue per recipient for email and SMS, but I need three things first, all required:

1. Rows: per flow or campaign: name, type (flow or campaign), channel (email or SMS), delivered count and attributed revenue. Why: RPR is attributed revenue divided by successful deliveries per row, so I need both numbers for each row, not only the $0.40 overall.

Facts used

Copperleaf Tea, send date 2026-11-05, segment 'Bought Earl Grey, no order in 90 days'. Avoid: 'cozy', emoji. - F1: "launch of Smoked Earl Grey, a new blend with lapsang souchong and bergamot" - F2: "50 g tin $14" - F3: "15% off the new blend with code SMOKY15" - F4: "ends 2026-11-09 at midnight ET" - F5: "Free shipping on orders over $40 (always on)" - F6: "brewing tip, 95 C water for 3 minutes" - F7: segment 'Bought Earl Grey, no order in 90 days'

Pairs

I can write Black Friday subject and preview pairs, but I need the final offer first. Both lines you suggested make claims I cannot back yet.

Required: 1. The confirmed discount. "Maybe 40-70%" is not settled, and "up to 70% off" is only honest if some items really are 70% off. Please confirm the range or the single discount, and what it covers (sitewide, a category, selected items). 2. Offer dates. When the sale starts and ends, with a time zone. Deadline lines need a real date.

1. Entry rule check

Your segment compared rule by rule with the Klaviyo example segment (Klaviyo help, sunset segment). Differences are questions, not errors.

Counted by script from the table:

Happy to write this series. Before I do, I need 4 things, because the emails can only say what you give me:

1. Platform. Which email platform you use. Why: the build steps I can give are sourced for Klaviyo only; for other platforms I will say they are not confirmed.
2. Segment. The entry rules as written in your platform (for example, emails received, opens and clicks over what period). Why: if the rules could catch recent buyers or openers, the wrong people get a goodbye email.

Exit Intent Popup Audit: harrowgate-knits.example

Tool: Klaviyo popup. Period: 2026-10-01 to 2026-10-28. 32 result rows (16 page patterns, desktop and mobile). Offer: 10% off your first order for your email.

1. Settings review

Sources: https://help.klaviyo.com/hc/en-us/articles/4413544555547 and https://developers.google.com/search/docs/appearance/avoid-intrusive-interstitials

2. Results by row

I can audit it, but I don't know the tool or the settings yet, and "annoying" needs numbers behind it. I need three things:

1. The popup's settings, as shown in the tool: timing, how often it shows again after someone closes it, devices, audience and which pages it targets. Your dev can screenshot or copy these, and the tool's name helps. If it is Klaviyo, I can use its setting names; for other tools I'll say the names are not confirmed in my sources.

llms.txt and FAQ build: PawWalk (pawwalk.test)

Source: the page descriptions you provided (fictional test data). I did not fetch the live pages, and you said fuller page text is not available right now, so every line below uses only the text you gave me.

1. llms.txt file in a code block

Only sections with pages are included (Product, Pricing, Optional). Docs and Policies are left out because no pages were given for them.

```markdown

PawWalk

I can build that. An llms.txt is an index of your most important pages with a one line description of each, so I need to know which pages and what they say:

1. Your site URL: the root domain, for example yourbrand.com. What name does the site use for itself?
2. Key pages: up to 20 URLs that matter most (home, product, pricing, docs, policies), with one line on what each is.

Optional:

Form Field Friction Review: 'Book a demo' on tallyroute.example

1. Field review

Form goal: book a sales demo. Marks come from the field_use you gave, compared with that goal.

2. Label and autocomplete checks

• Labels: W3C's labeling tutorial gives the wording "Labels need to describe the purpose" (https://www.w3.org/WAI/tutorials/forms/labels/). Q7 has only a placeholder, so it has no lasting label.

I can review your demo form, but I need two things before I start, so I don't guess which fields matter.

1. The fields. For each one: the label as it appears, the type (text, email, select and so on), whether it is required, and its autocomplete value if you know it. A paste of the form HTML or a screenshot of the editor works too.

Sources checked

Orbital Notes, standard property, range 2026-09-01 to 2026-09-30. Sources kept exactly as pasted. "Named assistant" means one of the five you listed (ChatGPT, Perplexity, Gemini, Copilot, Claude) matched to its source as it appears in your list.

Regex and test

Proposed condition for a channel named AI Assistants: session source matches regex

^(chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai)$

I can build an AI assistant channel for GA4, but I need your data first, and I will not change your property's default without your say so.

1. Source list (required). From Traffic acquisition, the session source and sessions for one date range, as a pasted table. Why: I test the channel rule against your real sources so it does not catch the wrong ones (for example, Google's help page example regex starts with '^.*ai', which also matches any source containing those letters).

Domains to configure

Your tag_info says the same GA4 measurement ID is on lanterntheatre.example and tickets.boxly.example, so the tag check passes for both. Google's page scopes the tag to "each domain that you want to include". pay.stripe.com is a third party payment page you cannot tag, so it is not configured here; it goes to unwanted referrals below.

Unwanted referral candidates

The three rows are not added together; I don't know whether the same visits appear in more than one row.

Hop tests

I can plan your cross domain setup, but I need a few things first so I don't configure the wrong domains.

1. Domains: every domain in the journey from your site to the checkout and back, for example [your site domain] and [your checkout domain], in order.
2. Hops: how a visitor moves from one domain to the next (a link, a button or a form that posts), and whether any step redirects. Redirects can strip the parameter that carries the visit across domains, so each one needs a test.

Measurement goals

• Revenue: class pack purchases (online sales).
• Leads: intro class bookings.
• Recommended event names were checked against GA4's Recommended events page, https://support.google.com/analytics/answer/9267735, opened 2026-10-29.
• Implementation: Google Tag Manager. PII rule (default): no name, email, phone or full address in any parameter. Your booking form collects name, email and phone, so none of those fields is sent.

Event plan table

Key events

I can draft the plan. To get the event names and key events right, I need a bit more, because "leads and stuff" could map to several different events:

1. What counts as success (required). Which of these are business goals: a demo request, a contact form lead, a newsletter signup, a purchase? Only goals become key events. Also, how does the business make money? That decides whether ecommerce events belong in the plan at all.

Rules per location

IP entries checked by script (Python ipaddress module).

You want Austin and the VPN kept apart, so each gets its own traffic_type value. The value names below are my suggestion; please confirm or rename them.

Which rows proceed: L1 and L2 (valid). Which row is held: L3 Agency gets no rule and no filter until you send a valid address (question 1). Rows: 3 (2 proceed, 1 held), counted by script.

I can plan this, but I need three inputs from you, and two parts of your request (the app, and "right now, permanently") change what I can do.

1. Office IP addresses (required): each location's name and its IP address or range. Why: Google's internal traffic rule matches on IP.
2. Existing data filters (required): name, type and state of any filters already in the property. Why: there is a limit of "up to 10 data filters per property".

Rows and shares

Tallgrass Outfitters, 2026-10-01 to 2026-10-28. Rule: share = row sessions / total sessions (52,000), by script. The two rows are kept exactly as labeled and are not added together: they come from different dimensions and may overlap.

URL tag check

Checking utm_source, utm_medium and utm_campaign (Google's URL builder page lists these, for example "utm_medium: Marketing medium").

Setup leads

Lead map

Leads, not proven causes.

Check list

I can find likely causes of your Unassigned traffic, but I need a few things first, and I will not change tags or settings until you approve.

1. The rows (required): the GA4 rows where session source / medium is (not set) or the channel is Unassigned, with sessions, and total sessions for the same range. Why: I keep the two labels apart and need the share of each.

GA4 Scroll and Video Engagement Plan: shiftly.example

Property: standard. Enhanced measurement on the web stream: Scrolls on, Video engagement off. Registered event-scoped custom dimensions: 14. Page list: 32 pages with September views.

1. Questions mapped

2. Video coverage by page

Classes: covered = YouTube with JS API on; not covered = YouTube with JS API off, Vimeo or self-hosted; no video; unknown. No page was unknown in your list.

3. Coverage counts

I don't have a login to your GA4 property, and I wouldn't change settings without your approval anyway, so I won't set anything up today.

Two things to know first: - GA4 enhanced measurement records a scroll only once per page, when a user reaches 90% depth. Google says scrolls fire when "a 90% vertical depth becomes visible". Scroll points at 25, 50 and 75 percent would need a custom event, and the Tag Manager setup steps for that are not confirmed in the sources I have.

GA4 weekly anomaly report: week of Oct 19 (Oct 19 to Oct 25)

Data: your fictional export (run 2026-10-29, v0.3.0), weeks labeled by Monday start date. All five weeks (Sep 21 to Oct 19) are complete; Oct 25 is before today (Oct 29). Key event: purchase. Threshold: more than 15 percent beyond the prior four week range.

1. Summary

• Organic Search purchases fell to 120 and revenue to $7,200, both about 24 percent below the prior four week low, while organic sessions were normal (8,050).

I can check. I only flag moves that are real against your recent history, so I need:

1. A GA4 export with five complete weeks: week start date (Monday), session default channel group, landing page, sessions, key events, and revenue if you track it. A GA4 exploration export works. I compare the latest complete week to the four before it; a week that has not finished yet is left out. If your weeks start on a different day, tell me which.

Category, services and description review: Juniper Street Bakehouse

Scope: the profile text and business facts you pasted. I can only propose category names from the picker list you saw, since custom categories are not allowed (Platform facts). Character counts include spaces, by script.

Category review

Services review

I can review your yoga studio's categories, services and description and propose a cleaner set. One flag first: categories are not keywords, and using them only to show up for more searches goes against the guidelines, so I propose categories that describe what the studio is. To start, I need two things first:

Metric table

Sprig Salad Bar, 3 locations. Periods: 2026-08-01 to 2026-08-31 (Aug, 31 days) vs 2026-09-01 to 2026-09-30 (Sep, 30 days). The lengths differ by 1 day, (31 - 30) / 30 = 3.3 percent, which is within 10 percent, so I compare them as given and note the day counts. Interactions = calls + website clicks + directions; no messages or bookings were given.

All 3 rows are sums of B1 + B2 + B3, for example Sep views 4,350 + 3,100 + 1,200 = 8,650 and Sep interactions 915 + 728 + 58 = 1,701.

Change table

I can read your profile numbers, but this comparison cannot be made as sent, and a few items are missing.

Required: 1. Two periods of the same length, with dates. "This week vs last month" compares about 7 days with about 30, so 900 views against 3,800 and 12 calls against 70 are not a drop by themselves. Please send two equal periods with exact dates, for example September vs August, or two back to back 28 day ranges.

Reviews sorted

Policy basis

Google's contribution policy lists conflict of interest, including "current or former employment" and competitors.

V2 is negative but describes a visit; the owner has no record either way, so I see no policy basis. V4 has no owner fact and no policy signal in its text. Google says "Do not report a review just because" you disagree with it, so I do not suggest reporting V2 or V4.

Counts

Report steps (held)

For V1 and V3, each HELD until you approve that review:

I can sort these reviews and prepare reports for any that break Google's policies, but I need a few things first, and two parts of the request I can't do as asked.

1. The reviews (required): for each of the 5, the reviewer name as shown, date, stars and full text. Why: I judge each review on its own text.

Review reply drafts: Brightside Family Dental (fictional)

Inputs used: business profile (Brightside Family Dental, general and family dentistry in Austin, a regulated healthcare business; tone warm and plain; sign-off "Dr. Lee and team"), contact route (call (512) 555-0142 and ask for Rosa, office manager), reply policy (staff first names allowed if the reviewer used them; no refunds, discounts or compensation), sensitive topics (treatment details, billing) and 4 reviews.

Happy to draft replies for all 12. Before I start, I need a few things:

1. The reviews. I can't read your Google Business Profile from here, so please paste each review with its star rating, the reviewer's name as shown, the review text and the date.

Google Business Profile Weekly Post Writer: Spoke and Chain Repair

Business: bicycle repair shop, not a hotel. Four weeks from Monday 2026-11-02. Every fact below is copied from your message; I added none.

1. Week plan

2. Posts

W1 (Update, 2026-11-02)

• Body: New at Spoke and Chain Repair: e-bike battery health checks. The check takes 20 minutes, and you can wait while we do it.
• Button: Link button to https://spokeandchain.example/book (button label not confirmed in the sources)

I can write the month of posts, but I won't write "best rated shop in town". That is a ranking claim nobody has given me a source for, and I don't write reviews, ratings or awards the owner hasn't supplied. If there is a specific rating or award you can point to, send the exact wording and where it comes from, and I'll ask you before using it.

To write the posts I need:

1. business: what kind of shop it is, and whether it is a hotel. Hotels can't create Offer posts, so this changes which post types I can use.

Location table

Clearwater Pool Service, campaign 'Pool Cleaning', last 90 days to 2026-10-27. Your report shows user location (where people were). Rules: CPA = cost / conversions; cost share = row cost / total cost; conversion share = row conversions / total conversions. Totals are sums of the six rows.

Clicks 620 + 210 + 150 + 18 + 95 + 60 = 1,153. Cost $2,480.00 + $820.00 + $640.00 + $70.00 + $380.00 + $230.00 = $4,620.00. Conversions 64 + 14 + 4 + 1 + 2 + 0 = 85. Shares are rounded to one decimal.

I can review your state results, but I need the report itself first. Right now I have impressions ("spend the most", "converts badly I think"), and exclusions should not rest on those.

Required: 1. The locations report. From Google Ads, per state: clicks, cost, conversions and conversion value, for one stated date range. Please also tell me whether it shows user location or matched location, because the two can differ.

What you told me

Conversion actions to create

Google says calls from ads, website calls and imported calls "require Google forwarding numbers", and that these numbers are available only in the countries on Google's list. I have not checked that list for the United States; it is not in my saved sources (question 1).

Call length choice

Happy to plan it. I'll write the plan only; nothing gets created or changed in your Google Ads account until you approve it. To pick the right call conversion actions and a minimum call length for a law firm, I need five things:

1. Where calls come from: call assets or call only ads, a phone number on your website, tap to call on mobile, or a mix. Each maps to a different call conversion type, and a tap on a mobile site number counts only the click, not the call.

Drop confirmed

Lumen Learning Co. Today: 2026-10-29. You said conversions fell from Oct 20, and the daily rows agree: both campaigns go from conversions every day (Oct 13 to Oct 19) to 0 per day (Oct 20 to Oct 28). Drop date used: 2026-10-20. Window: 14 days before it (Oct 6 to Oct 19) plus the drop date.

Changes in window

Change types are as exported.

Changes in window: 4 (C1 to C4).

Before and after

I can find what changed, but I need a few inputs first. With view access you can still get most of them.

What I need (required): 1. The change history rows. If you can open Change history in the Campaigns menu but can't download it, filter to the right dates and copy the table (date and time, user, campaign, change type, old and new value), or ask an account admin to export it. Change history goes back 2 years, so last month is covered.

Google Ads Wasted Spend Audit: Rapid Rooter Plumbing (fictional)

Scope confirmed: residential plumbing repair, Tampa only; out of scope: new construction and commercial work. Target CPA: $60 for every campaign. Lookback: last 30 days. Data used: the search terms report (6 rows, with ad group and conversion value columns) and the campaigns report (3 campaigns). No keywords report was provided. Click threshold for judging a keyword: 30 (default). No connector is in use; nothing has been changed in the account.

I can run that audit. To find real waste I need four things first:

1. Search terms report for the last 30 days (CSV export) with campaign, ad group, match type, cost, clicks and conversions, plus conversion value if you track it.

Inventory counts

Container GTM-ABC123 'tidepool.example'. Live version 41. Your input says the workspace has 2 unpublished changes; review them before any publish. Names are exactly as exported.

Counted from your lists.

Duplicates

Broken items

Unused items

V1 'DLV - transaction_id' is used by T3. V2 'DLV - value' is used by T3 and T4.

Paused tags

I have not recommended deleting T8. Is it paused on purpose, or should Google Ads conversion tracking be on?

I can audit the container for duplicates, but I need its contents first. I also cannot publish anything: publishing needs your explicit approval of a named version, and you do it yourself.

What I need (required): 1. The tag list. For each tag: name, type, key ID (for example the G- measurement ID) or event name, firing triggers, blocking triggers, and whether it is paused. I look for two tags sending the same ID and event on overlapping triggers, so IDs and triggers are essential.

Link review

The Ledger Desk, 'How to Close Your Books in 5 Days'. Payment: "the author's company paid $300 for publication." Google asks that paid placements be qualified: Google says to mark "advertisements or paid placements" with rel="sponsored". The input does not name the author's company, so the rel column for closewise.example links is conditional on Closewise being the payer (question 1).

I can review the guest post and tell you whether to accept, revise or decline it. I need three things first:

1. The draft (required): the full text, with each link's anchor text and URL. The draft did not come through with your message. Why: I check every link and every statistic, so I need the real text, not a summary.
2. Your guidelines (required): the publication's guest post rules. Why: "good" here means it meets your rules, and I mark each rule Met, Not met or Cannot tell.

Templates checked

3 templates from fernhill-garden.example, headings numbered in page order exactly as you gave them.

Findings

I can audit the heading structure, but I need the headings themselves first. I will not fetch or guess them.

1. Outlines (required). For each template you want checked (for example one blog post and one product page), paste the headings in page order as "level | text", like: h1 | [your page's main heading] h2 | [next heading on the page]

Your question

"which channels should get credit for new contacts from our demo request form this quarter? Leadership thinks paid social is overrated." You also asked about revenue attribution.

Edition check: you are on Marketing Hub Professional. HubSpot lists Deal Create and Deal Revenue Attribution as "Marketing Hub Enterprise accounts only", so I cannot spec a revenue report for your account. Contact Create Attribution fits your question as asked.

Report spec

Interaction types to check

Yes. To spec a report your HubSpot account can actually build, I need four things:

1. Your Marketing Hub edition (Professional or Enterprise). "What's driving revenue" points to deal revenue attribution, and HubSpot offers the deal create and deal revenue data sources to Marketing Hub Enterprise accounts only. On Professional, the closest report is contact create attribution, and I'll say so rather than spec one you can't build.
2. The conversion to credit: contact created, deal created or deal revenue.

Images checked

Your notes: I3 shows the text 'Handmade in Maine' over a photo of mugs; I4 is a decorative divider; the LCP element on /shop/mugs is "the hero image".

Findings

Not flagged: I2 (descriptive filename and alt, lazy but not in the first viewport) and I4 (decorative, so empty alt fits the W3C alt decision tree, under which a purely decorative image may have empty alt (my paraphrase)).

Why these matter, from the saved sources:

I can audit your image SEO, but not from a description of the site, and not for thousands of images at once without a list. I need:

1. Image list (required). Per image: the page it is on, filename, alt text, whether it is an img tag or a CSS background, its loading attribute, and whether it is visible on first load (first viewport). A crawler image export plus your notes on first viewport works. If thousands is too many, start with your key templates, for example one product page and one category page.

Targets and phrases

Link suggestions table

Ordered by target, then by source page inlink count, highest first. Pages with more internal inlinks pass more internal authority; that is the ordering reason, not a measured effect.

No new sentences are needed: every anchor appears verbatim in your source text. Each anchor text points to only one target.

Links per target

Total = 2 + 2 = 4, which matches the 4 rows in the table.

I can build this link plan, but two required items need an answer first, because both change which links I suggest:

1. Which pricing page is the target? "Pricing stuff" matches two URLs in your inventory: https://brightpath.example/pricing (40 inlinks) and https://brightpath.example/pricing/enterprise (3 inlinks). Tell me one, or both with a keyword for each. The anchors differ for each, and the same anchor can't point to both.

Split queries

Site: https://www.hearthbook.example. Search Console, 2026-07-01 to 2026-09-30. Rules: share = page impressions / total query impressions (summed from the rows); a query is split when 2 or more pages each have at least 20%. Minimum query impressions: 200 (every query here qualifies).

Split queries: 3, counted from the table.

Page pairs

Actions

Pair A: Merge.

I can check whether those posts really compete, but I need data that shows how Google splits the queries between them, plus what each post is for.

What I need (required):

Intent labels

Cluster table

Rule: cluster volume = sum of the member volumes listed in the row. Primary keyword = highest volume member. Ranked by cluster volume, highest first.

Total volume across the 5 clusters = 4,500 + 2,490 + 1,870 + 1,110 + 880 = 10,850. All volumes are exact figures from your Semrush export; none is approximate.

Page mapping

Happy to cluster these for Paws and Suds. Before I start, one required item is missing and it changes the output:

1. Your existing pages (required). I need the list of URLs on https://pawsandsuds.example and the topic each one covers. Each cluster has to map to one page, and without the list I can't tell which clusters already have a home, which need a new page, or where two clusters would compete for the same page. "Whatever pages make sense" would mean guessing your site structure.

Page summary

• Goal: start a free trial. Primary CTA label: "Start free trial".
• Sections counted from your copy: 7 (Hero, Logos, Features, Proof, Pricing, Form, Footer).
• CTAs counted: 2 buttons with different labels: "Book a demo" (Hero) and "Start free trial" (Form).
• Form fields counted: 6 (name, email, phone, business name, number of staff, how did you hear about us).

I can tear this down, but two required items are unclear and they change every recommendation:

1. One primary goal. You wrote "signups or demos, whichever." The page has one main job, and every fix depends on it. For example, a 9 field form might make sense for demos and be far too long for signups. Which action matters most, and what is its button label?

Page table

Site: https://www.trailmixbudget.example. Search Console, web search, last 28 days to 2026-10-27. Rule: CTR = clicks / impressions, recomputed for every row. Minimum impressions: 1,000.

Average position is the average of your site's topmost position for each impression. It is an average, not a fixed rank.

Band medians

Priority list

Rule: clicks at stake = impressions x (band median CTR minus page CTR), floored at 0. These are estimates of the gap to your own band median, not forecasts.

I can rank these, but the data you sent cannot be compared yet. Four things are needed first, and each one changes the ranking.

1. One date range for every page (required). Some rows cover the last 3 months and some the last 28 days. CTR and impressions over different spans are not comparable. Please export all pages from Search Console for one range. Default if you do not choose: the last 28 days.

Data check

Period: 2026-09-01 to 2026-09-30. Business: Orchard CRM. All arithmetic below was run by script.

• Rows: 8 paths.
• Total conversions = 40 + 25 + 10 + 30 + 20 + 15 + 10 + 5 = 155.
• Total revenue = 12,000 + 9,000 + 4,000 + 7,500 + 5,000 + 6,000 + 3,500 + 2,000 = $49,000.
• Channels: Paid Search, LinkedIn Ads, Organic, Email, Referral. Names are used exactly as given.

Per path credit

I can run first touch, last touch and linear on your paths. Three things need sorting out first, because they change the totals:

1. A missing conversion count. The row "Organic > (direct)/(none)" has no number. A path with no count can't get credit. What's the count, or should I drop that row?

Merchant Center free listings check: Bramblecraft Candles

Target country: United States. Account created directly in Merchant Center. Every change below is HELD for your approval. Free listings status is permission to show, not a guarantee: Google's page says it "doesn't guarantee that your products will be shown" (https://support.google.com/merchants/answer/13889434?hl=en).

1. Product gaps

I can check your soap products for free listings, but I need 3 things first:

1. Product rows: id, title, link, image_link, price, description, availability, condition, gtin, mpn and brand for each soap. I check each attribute against the scope Google gives it, so a blank value only counts as a gap where the rule applies.
2. Target country: where you sell. Shipping settings are required in some countries, and the check depends on yours.

Flags

Checked against Google's title rules: no promotional text, "Don’t use capital letters for emphasis", no extra white space, and "Add the distinguishing details of each variant".

F4 note (my own, not a Google rule): "UV-Resistant" is not in F4's attributes, so I left it out. If it is a real product fact, tell me its source and I will add it back (question 1).

Rewritten titles

Order: brand, product type, key attribute, variant (your default).

Character counts

I can rewrite your Google Shopping titles, but I need the product data first, and two parts of the request go against Google's title rules.

1. Products (required): for each, the id, current title, brand, product type and variant attributes (color, size, material and so on). A feed export works. Why: I build each title from the product's own attributes, and variants need their distinguishing details.
2. Order of details (optional): default is brand, product type, key attribute, variant.

What I can't add:

Facts used

All facts are quoted from your input. Character counts below were checked by script and include spaces and punctuation.

Not used: "fresher than supermarket beans". Your audience note says these buyers want fresher coffee, but the facts contain no comparison with supermarket beans, so no line makes that comparison. The banned words "cheap" and "best" appear nowhere.

Happy to write these. A few items need confirming first, because Meta ads can only say what you can back up and what the landing page shows:

1. Landing page URL (required). Where do the ads send people? Every line has to match that page.
2. Brand name and offer details (required). What is the subscription called, and what does it cost? Right now I have no price, bag size, delivery schedule or offer to write from.

Reference

You said the website and Business Profile agree on this:

Normalizing is for comparison only: case, punctuation, Street vs St, Suite vs Ste, and phone digits. Nothing else is treated as the same.

Field comparison

Notes: - C2 name adds "Best Denver Dentist". On Bing Places this is my own note, not a Google rule: Google's naming guideline applies to the Business Profile, where the name should be the one "used consistently on your storefront, website" and stationery.

I can check your name, address and phone across listings, but I need two things before I start, and the move makes the first one important.

1. Reference NAP (required): the business name, new address and phone exactly as they should read. Tell me if your website and Google Business Profile both show this now. Why: every listing is compared against it, and if the two disagree I need you to pick one.

Newsletter Performance Review: 8 sends, Aug 4 to Sep 22, 2026 (fictional data)

Goal: sponsor revenue, with sponsors paying per click. So clicks per send and click rate are the metrics that matter most here. Open rate is shown for completeness only: privacy features such as Apple Mail Privacy Protection preload emails and record "opens" that no human made, so opens are inflated and not a reliable success measure.

Glad to dig in. To give you an honest read instead of a guess, I need two things:

1. Per send stats for your last 8 to 12 issues, ideally a CSV with send date, subject, delivered, unique opens, unique clicks and unsubscribes. If your platform exports total rather than unique opens or clicks, tell me. With fewer than 8 sends, any trend I report won't be reliable.
2. What the newsletter is for: sponsor revenue, product signups, retention, or something else? It decides which metric counts as "doing well."

Coded responses: "What almost stopped you from signing up?"

Limits first: this rests on 36 responses from one month, so the shares below describe these respondents only, not all sign-ups. I read all 36 responses before drafting codes (the default is at least 10%, and all when under 50). One coder (me) assigned every code; no agreement check has been run yet.

Codebook

Other: no response needed it; every response fits at least one code.

Coded responses

2 responses carry 2 codes: R4, R26.

Code counts

I can code your churn comments into a codebook with counts and quotes. Before I start, I need two things first:

1. The question (required): the exact question text these answers respond to. Why: I code one question at a time, and the wording decides what a code means.

What I compared

I compared the 15 paths your crawl reached through internal links with the 18 paths from your sitemap and Search Console (period: last 3 months, as you said). A path counts as an orphan only when it is in other_urls and not in crawl_urls. I matched paths exactly; none needed trimming. I did not check whether any page is indexed or has external links: your lists do not say, so I do not claim either.

I can check this for the dental practice site, but I only call a page an orphan when two lists prove it, so I need these before I start:

1. crawl_urls: the URLs a crawl reached by following internal links from the homepage, with the crawl depth of each. This is the linked side of the comparison.

Date table

Site: https://docs.relaybase.example. Audit date 2026-10-29. Substantive means a change to facts, steps, prices or more than a paragraph (default threshold); typo and layout fixes do not count.

Flags

• P2 Unlabeled: the visible date 2025-01-15 has no label. Google's byline date guidance asks for a visible date labeled with text such as "Published" or "Last updated".

I can't do this as asked, and I need some inputs before I can help with dates on these posts.

Templates checked

Pages in series = total items / items per page, rounded up, computed by script. For T2 and T3 these pages do not exist as URLs today; the number is how many a paginated version would need.

Check results

Five checks per template. Check (e) is a Note, not a Fail, because Google only says to "consider linking from all individual pages in a collection back to the first page".

Test question

Which opening hook, built from a real customer quote, gets the most purchases per dollar for the Driftwood Sleep weighted eye mask ($38, free shipping on first order) on Meta?

Fixed and variable elements

• Variable (the only thing that changes): the hook, meaning the first line of the video and the first line of the primary text.

Thanks. Those three reviews give me good hook material for Summit Bar. Before I build the plan, three required items need a clear answer, because the budget math and the cells depend on them:

1. Budget amount and period. "Around 2-3k" is a range with no period. Is that $2,000 or $3,000, and is it per day, per week, or for the whole test? I split the budget evenly across cells and show the per cell math, so I need one number and its period.

Facts per Pin

Pins

P1 (landing page https://thistlethyme.example/red-lentil-soup) - Title: Red Lentil Soup with Lemon, One Pot and Ready in 30 Minutes - Description: Red lentil soup made in one pot with red lentils, carrots, cumin and lemon. Ready in 30 minutes, vegan, serves 4. An easy vegan soup recipe.

P2 (landing page https://thistlethyme.example/sheet-pan-chicken) - Title: Sheet Pan Dinner, Chicken Thighs with Baby Potatoes and Rosemary

I can write Pinterest titles and descriptions for your 20 products, but I need a few things first.

1. Per product (required): an image description, the landing page URL (https://) and the facts on that page, such as material, size, price or what it is for. Why: Pinterest asks that Pin text match the landing page, so I only use facts you give me.
2. Keywords (required): the search terms you want each Pin, or the set, to target. A Pinterest Trends export works.

Status groups

Meadowlark Wellness, targeting United States for all campaigns, no certifications (your input). Status and Policy details kept exactly as shown.

Per item actions

Notes per item:

I can sort your limited ads and prepare appeals, but I will not appeal all of them at once, and I need the details first.

Why not appeal everything: Google's help says an Eligible (limited) ad "can run, but not in all situations", and the limit is often a correct restriction (Google lists limits such as location, device and age). An appeal only fits when the label is wrong for that ad. Appeals are also a step you approve one by one.

What I need:

Positioning statement

For owners of a single cafe who ran out of stock before Tallyhaus [E2] and want stock counts that take minutes [E4], Tallyhaus is an inventory and reorder app that counts stock from shelf photos [product facts, E4]. Unlike spreadsheets and clipboards [E1], it suggests reorder quantities from the last 8 weeks of sales [product facts, E5].

Target segment

Segments use the labels exactly as they appear in your evidence. No labels are merged.

I can draft this memo for Tallyhaus. Positioning rests on evidence, though, and three required pieces are thin or unconfirmed:

1. Customer evidence (required). "Customers love the photo counts" is a summary. I need the actual words: quotes, interview notes or survey answers, each tagged with who said it (for example "owner, single cafe"). The memo cites a line for every claim, and the target segment comes from counting who said what.

Post table

Juniper Yoga Studio, Instagram, last 8 weeks. Times are America/Chicago and the audience is in America/Chicago, so no conversion is needed. Blocks (default): Morning 06:00 to 10:59, Midday 11:00 to 13:59, Afternoon 14:00 to 17:59, Evening 18:00 to 22:59, Late 23:00 to 05:59. Rule: engagement rate = engagements / impressions.

Posts: 12.

Cell table

Rule: cell rate = sum of engagements / sum of impressions. Cells need at least 3 posts.

Check: 4 + 3 + 3 + 1 + 1 = 12 posts.

Recommended slots

I can find your best slots from your own history, but I need post level data with clear time zones first. A generic "Tuesday at 10 am" may not fit your audience; your own numbers will tell us.

What I need (required):

Pacing summary

Kestrel Home Insurance Agency, October 2026. Monthly budget: $12,000.00, hard cap. Numbers run through 2026-10-20, end of day; you marked the day complete, so I count it as elapsed. - Days elapsed: 20 (day of month of 2026-10-20). Days remaining: 31 - 20 = 11. - Month to date cost: $4,100.00 + $2,300.00 + $1,250.00 = $7,650.00. - Remaining budget: $12,000.00 - $7,650.00 = $4,350.00.

I can work out new daily budgets for the rest of the month, but I need a few details first, because each one changes the numbers.

Required: 1. Per campaign: average daily budget and month to date cost for Brand, Generic, Shopping and the Display campaign. "About 15k" in total is not enough: I project each campaign from its own numbers and then add them up.

PPC Landing Page Message Match Review: PeoplePal (fictional)

Conversion goal: booked demo. Ad groups reviewed: 2.

What I scored from: the page snapshots you pasted (desktop view, captured Oct 28). I did not load the pages; the URLs are on a .test domain, which is not a live site. Source for both pages: pasted copy. Review date: 2026-10-29. Your snapshots covered all five elements (headline, subhead, CTA, offer, proof), so no total needed scaling.

Score table per ad group

I can check that. I need two things to score each ad against its landing page:

1. Your ad groups: for each one, the name, top 5 keywords, ad headlines, any offer the ad mentions (price, trial, discount), the final URL and the cost for the period. A Google Ads export with those columns works.
2. Your conversion goal: what counts as a conversion, for example a booked demo, trial signup, purchase or call?

Angles by journalist

Fact sheet lines are numbered 1 to 5, as you supplied them. Because the news is under embargo, the first pitch only asks each journalist to agree to hold it; the figures, quote and assets go in a details email after they agree.

The pitch to Tom does not suggest the redesign caused the ranking. The fact sheet does not say that.

Pitch drafts

I can draft these pitches, but some things need sorting out first. Also, I don't send anything. I save drafts for you to approve one by one, so "send them out Monday" will be your call once the drafts are approved.

What I need: 1. The approved quote (required). "Something like 'reliability matters most'" isn't approved wording. I quote the CEO word for word, so please send the exact quote your CEO has signed off, with their name and title.

What I used

Your offer (private in-person piano lessons in Columbus, at the studio or in the student's home within 10 miles), your not_offered list, 7 planned keywords in one Search campaign, and 40 Keyword Planner ideas with average monthly searches. Nothing was added to any account.

Idea review

Negative lists

Happy to build this before the bakery's ads go live. A negative keyword is only right if it blocks searches for things you don't sell without blocking searches you want, so I need these from you:

1. offer: what the bakery sells, where (city or delivery area), to whom, and whether you see yourself as budget, mid or premium, in your words.

Launch summary

• What: Shared Budgets in Penny Planner. Two people can share one budget and see each other's spending in real time.
• Who gets it: free for current Plus subscribers. Plus is $6.99 per month.
• Where: iOS and Android.
• Launch: Tuesday 2026-12-08, 9:00 AM CT. Tier 2.

Timeline

Rule: task date = launch date minus lead days. A date that falls on a Saturday or Sunday moves to the Friday before. Every date was checked by script.

I can build the Shared Budgets plan, but every date is worked back from the launch date, so I need these first:

1. Launch date (required). "Early Q1" isn't a single date. Give me one calendar date and time zone, for example 2027-01-12, 9:00 AM CT. Even a tentative date works; we can shift it later.

Keywords reviewed

Account: Fernhill Dental, Austin. Last 30 days. Cutoff: Quality Score 5 or lower. Quality Score is a keyword level diagnostic on a 1 to 10 scale. It is not an input to the ad auction, so these fixes target what the components measure, not the score itself.

Not reviewed: K3 'implant dentist' (QS 7, above the cutoff). Reviewed: 4 keywords. Cost of reviewed keywords: $1,240.00 + $860.00 + $940.00 + $310.00 = $3,350.00.

Component counts

I can diagnose these, but I need the keyword data first. A 3 or 4 doesn't tell me what is wrong; the three components behind it do.

What I need (required):

Post table

Northwind Freight Software, LinkedIn and Instagram, 2026-07-01 to 2026-09-30. Goal: engagement rate. Rules: engagement rate = engagements / impressions; click rate = clicks / impressions. Labels are exactly as exported.

Posts: 12. Note: LinkedIn and Instagram may count "engagements" differently, so compare platforms with care (see Questions for you).

Group rates

Rule: group rate = sum of engagements / sum of impressions for the group's posts. Groups with fewer than 3 posts show "too few posts".

I can find out what happened last quarter, but I need post level data first. Monthly totals can't show which posts did well or badly, or why.

What I need (required):

Customer and order counts

All 45 orders fall inside the range and every order has a customer ID (checked by script).

Cohort table

I can work out your repeat purchase rate from this export, but two things need settling first, and both change the answer.

1. Date range (required). You are not sure what period the export covers. Please give the first and last order date to include. Why: a customer counts as repeat only if their 2 or more orders fall inside the range, so a different range gives a different rate. I will not pick one from the dates in the paste.

Returns by reason and product page gaps: Saltmarsh Supply, Aug 1 to Sep 30, 2026

Limits first: this rests on 33 return rows (33 units) across 4 products. Each product has between 4 and 12 returned units, so a single return moves its numbers noticeably; treat the figures as directional. Reason labels are kept exactly as exported (7 distinct labels). Counts and divisions are by script.

Returns by product

Rule: returned units / units sold in the same period.

Reasons by product

Page gaps

I can tally your rain jacket returns by reason and map the top reasons to gaps in the product page. Your message has no returns data, so I need four things first:

1. Returns (required): rows of product, variant, quantity returned and return reason as recorded, plus any customer comment. Why: I sum by product and reason by script and keep your reason labels exactly; I only state the recorded reason and the customer's own words, never a guessed one.

Program rules

Bluebird Auto Glass, Denver. Service: windshield replacement and chip repair. Sender: Marcus, owner. Trigger: job marked complete in your shop software. 1. Ask every customer whose job is marked complete, not only the ones you expect to be happy. No survey or "rate us first" step that sends only high ratings to Google: Google's Maps content policy prohibits discouraging negative reviews or selectively soliciting positive ones.

I can write review request texts for you, but not the program as described: two parts break review rules, and some inputs are missing.

What I will not do, and why: 1. Ask only people who rated you 4 or 5. That is review gating. Google's Maps content policy prohibits selectively soliciting positive reviews, so the program asks everyone who meets the trigger.

Data summary

• Business: Lumen Lane bike lights. Source: Trustpilot export, October 2026. Period: all rows supplied (2026-10-02 to 2026-10-28).
• Reviews: 10. Rated: 10. Unrated: 0.
• Average rating = 3.6 (worked out under Rating counts).

Rating counts

Average rating = 36 / 10 = 3.6. Checked by script.

Review tags

Theme table

Share = reviews tagged / 10 reviews. A review can carry more than one theme, so shares add up to more than 100%.

I can summarize these, but a few things in the paste change the counts:

1. Two products are mixed together. The paste covers Lumen Lane lights and the Lumen Lane Helmet Cam. Should the digest cover the lights only, the helmet cam only, or both in separate sections? Mixing them would blur the themes, for example "stopped recording" next to "super bright".

Groups and rules

Site: https://www.oakandloom.example. Crawler tested: Googlebot. File size: 1.2 KB, under Google's 500 KiB read limit.

Group choice: the file has no Googlebot group, so Googlebot follows the * group (R1 to R9). Googlebot-Image (Google's image crawler token, per https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers) matches its own group (R10), and only one group is valid for a crawler, so it follows R10 only.

URL test table

I can check this file, but I need a list of URLs to test it against. Whether a rule is a problem depends on which paths it hits.

What I need (required):

Responsive Search Ad: TidyBooks, Shopify bookkeeping ad group (fictional)

Happy to write them. Every line has to be backed by something real, so I need three things first:

1. Your landing page: the URL the ads will send people to, or paste the page copy. I only write claims that appear there.
2. Your keywords: the main searches you want these ads to show for.
3. Your offer and proof: any price, promo or free trial, plus proof you can back up, such as review ratings, customer counts or awards. If it's all on the landing page, just say "same as landing page."

Parse result

• The JSON-LD parses: yes (checked by script). No syntax errors.
• @context https://schema.org, @type Product. This matches the page type you gave.
• Property list sources: Google Search Central, "Product snippet structured data", https://developers.google.com/search/docs/appearance/structured-data/product-snippet, and "Review snippet structured data", https://developers.google.com/search/docs/appearance/structured-data/review-snippet, both opened 2026-10-29.

I can check this, but I need three things first. One blocking error is already visible: the block does not parse, because there is a trailing comma after the aggregateRating object, just before the closing brace. Search engines ignore markup that does not parse, so fix that first.

What I need:

Network split

Cobalt Legal Intake, 2026-09-01 to 2026-09-30, target CPA $120. Network labels kept exactly as in your report. Rules: CPC = spend / clicks; CVR = conversions / clicks; CPA = spend / conversions (n/a at 0 conversions); shares are within each campaign; CPA change is versus that campaign's Google search row. All computed by script.

Flags

I can check whether search partners and the Display Network are hurting your CPA, but I need numbers per campaign first. Turning them off everywhere without that could also remove conversions that are cheap for you.

1. Campaign settings (required): for each Search campaign, its name and whether "Include Google search partners" and the Display Network are checked.

Promote

Account: Pebble & Paw, Search campaign 'Generic' report, last 30 days to 2026-10-27. Conversion action: quote form submit. Thresholds: promote at 2 or more conversions; negative at $120.00 or more cost with 0 conversions. Target CPA: $60.00.

Google prioritizes an exact match keyword that is identical to the search term over other keywords, so each term will serve from the ad group named.

Negative candidates

I can sort these into keywords to add, negatives and a watch list, but a few required items are missing, and they change which list each term lands in.

1. Cost and conversions per term (required). "Spent a lot" (roof repair cost) and "some" (metal roof) aren't numbers. For every term I need clicks, cost and conversions from the search terms report. 'roofing jobs' has clicks but no cost or conversions. Is "3 leads" for 'roof repair near me' the conversion count in Google Ads, or leads counted elsewhere?

Self-reported attribution: Kettle and Crumb, Sep 1 to Sep 30, 2026

Limits first: this rests on 40 survey answers from 112 orders in one month. Answers are what customers remember about first hearing of you; they are not tracked conversions, and shares on this few answers move a lot with each answer.

Survey question

Your current options are not randomized. Proposed question, HELD until you approve publishing it:

How did you first hear about us? (channel order randomized each time) - Instagram ads

I can count your survey answers by channel and compare them with spend, but I can't say whether to move budget from this message, and survey shares are what customers remember, not tracked revenue or ROAS. To start, I need three things first:

1. Channels (required): every channel you are active in, in your own labels. Why: I count answers per channel and check the options the survey offers.

Content brief: "how to calculate customer acquisition cost"

Site: fundmetric.test (fictional finance dashboard for startups) Audience: seed stage founders who are not finance experts Length: about 1,800 words (your target) Keyword volume: unavailable (no Ahrefs or Semrush access this session); not estimated. Sources: your pasted SERP notes and People Also Ask questions. I did not open the result pages myself.

Intent

I can build that brief. "Email marketing" is very broad, so a few details decide the whole angle:

1. Your site: what's the domain, and what does the business sell?
2. Your audience: who is the reader, and how experienced are they? A solo shop owner sending a first newsletter needs a very different piece from a marketer comparing tools.

Facts per city

Rule: a page is planned only when a city has at least 3 unique, verifiable facts (your threshold). Each fact is counted once.

Pages planned: 3 (Boise, Meridian, Eagle). Held: 2 (Kuna, Star). Check: 3 + 2 = 5 cities served.

Pages to build

Page outlines

Each section cites the city fact it uses. The general service description is one short section per page, and it differs only through that city's facts.

Boise (/boise-plumber)

I can plan city pages, but not write a page for every town within 50 miles. Pages that only swap the city name are thin, near duplicate "doorway" pages, which search engines treat as spam. Each page needs real facts about that city. Here's what I need:

1. Named cities (required). "Everywhere within 50 miles of Boise" isn't a list. Which cities or towns do you actually serve?

Hook options

Hook beat budget: 3 seconds x 2.5 words per second = 7.5, rounded down to 7 words. Each option is within it. Word counts below are whitespace tokens, computed by script on the final text.

Chosen for the script: option 3, because it opens on the exact measure people need. The hook appears only in the hook row below.

Word budget

Rule: total voiceover words = 30 seconds x 2.5 = 75. Words per beat = beat seconds x 2.5, rounded down.

Script

Happy to script this for Hearthwell. Four required items are open, and each one changes the script:

1. One length in seconds. "15-45 seconds" is a 3x range. At 2.5 words per second, that's anywhere from 37 to 112 words of voiceover, so the beats and the number of steps that fit are very different. Pick one number.

Four week social calendar: Little Fern Bakery (fictional)

Happy to plan next month. I need three things first, so the calendar fits your business:

1. Your brand: what do you sell, and to whom?
2. Platforms and frequency: which platforms, and how many posts a week on each? If you're not sure, I'll suggest a minimum you can keep up.
3. Your main goal for the month: for example more sales of one product, event signups, preorders or followers.

Helpful but optional:

1. Row filter

Band: 8 to 20 (default, inclusive). Min impressions: 100 per row (default, my working cut, not a Google rule). Brand terms: crumbhouse (whole word, case ignored). Source: Search Console Performance, web search, 2026-08-01 to 2026-10-28.

Excluded rows:

2. Kept queries

Ranked by impressions, highest first. CTR = clicks / impressions, computed per row.

3. Pages to work on

Per page sums of the kept rows only, ordered by impressions. Page CTR = summed clicks / summed impressions.

I can't pick keywords from memory: every query, page and number I give you has to come from your own Search Console data, and I can't promise any query reaches page one. Changes are tests you measure in the same report later.

I need 2 things before I start:

Technical SEO Crawl Triage: shop.example (fictional test data)

Inputs used: crawl export (8 URLs, with redirect target and meta description columns), site type ecommerce, priority sections /products/ and /collections/, dev capacity 6 hours next sprint. Not provided: Search Console Pages report, so I could not cross check crawler indexability against Google's view. Columns not in the export: H1 and robots directives, so missing H1 and robots blocks could not be checked. No live crawl or tool was used.

I can help track down the indexing problem. I can't crawl example-store.test from here, so I need two things before I start:

1. A crawl export as a CSV from Screaming Frog or any crawler, with these columns per URL: status code, indexability, canonical, redirect target, title, meta description, word count and inlinks. If your crawler leaves some out, send it anyway; I'll list which checks I couldn't do.

Event map

E3 is sent by both the pixel and the Events API because you said the backend can send CompletePayment when the order is paid. E1 and E2 are pixel only, since no server source was named for them. TikTok recommends "Events API integration with your existing Pixel", so adding E1 and E2 to the backend later is an option, not a requirement here.

Deduplication rule

For E3 only:

I can plan the TikTok pixel and make purchases count once, but I need a few details first, and I won't install anything until you approve the plan.

1. Site (required): the URL and what it runs on (Shopify, custom code, other). Why: it decides between a partner setup and adding code by hand.
2. Events (required): which actions to measure and where each happens, for example "purchase on the order confirmation page". Why: each event needs a clear trigger.

Mention triage

Brand: Tallyfox (variants Tallyfox, TallyFox, tallyfox.app). Source: your Ahrefs alerts, October 2026. Skip list: spendly.example (no mention row is on it).

Ranking rule: contact known before contact unknown, then exact topic page before home page. All three pursued rows match an exact topic page.

Outreach drafts

Body word counts were computed by script. They exclude the subject line and the URL; the limit is 120.

M1, to Dana Ortiz, dana@smallbizweekly.example

I can help Quillbase reclaim unlinked mentions, but I can't do it the way you described, and I'm missing the inputs I need to start.

What I won't do: - Offer a free year of Pro for a link. Google's spam policies treat links exchanged for money, products or services as link spam unless the link is marked rel="sponsored" or rel="nofollow". A link traded for Pro would put both sites at risk, so the drafts will ask politely, with no incentive.

Parsed parameters

Site: https://northpeak.example. Convention: the default (lowercase; hyphens between words; utm_medium from cpc, paid-social, email, social, referral, affiliate, display). Parsing and counts were done by script. Encoded values are decoded for display (%20 and + are spaces).

Issues by row

Each issue type is counted once per row. Medium is checked against the allowed list after lowercasing, so a case problem counts only as uppercase.

Issue counts

Rows showing each issue type:

I can audit these three links. First, one decision changes every corrected URL:

1. Underscores or hyphens? Your rule says underscores, and the agency uses hyphens. GA4 doesn't require either one. It records whatever value you send, and it treats different spellings ("paid_social" and "paid-social") as different values. It also treats different cases as different ("Meta" and "meta"). So the right answer is to pick one and use it everywhere. Which wins, your rule or the agency's?

Value Tier Lookalike Plan: Juniper Loom

Value means: total net order value per customer over the last 24 months, in USD. Full file: 4,820 customers. Goal: prospecting. I worked only on row ids and values; the email column stays redacted.

1. Value column check

• Meta's customer list guidelines say lists should "include at least 1,000 customers" and value columns must be numeric "without any currency characters" (https://www.facebook.com/business/help/2082575038703844).

I can plan it, but the list comes first: a value-based lookalike is seeded from a customer value audience built from the list, so I'd be setting up something with no source. I need three things:

1. What the value number means, in your words: order value, lifetime value, items bought, or something else. Meta lets you choose, but it has to be the same meaning in every row.
2. How many customers are in the full file. Meta's minimum is in the table below; if the file is under it, I stop before tiering.

Pages checked

Site lumen-hr.example, CMS Site plan. Auto-generate sitemap: Yes, so the Sitemap indexing toggle can be used (Webflow says you can only disable indexing with the toggle "if you've enabled your site's auto-generated sitemap"). Title characters counted by script, including spaces.

Findings

1. Index conflict, P4 /thank-you: toggle is on, you want it out of search.
2. Index conflict, P5 /careers: toggle is on, you want it out of search.

Before I audit, I need a few things. I have no page data yet, and "important pages" and "junk" depend on your call, so I will not guess.

1. Page list (required). For each page: slug, type (static, CMS template or CMS item), title tag, meta description, and whether Sitemap indexing is on or off. A copied table is best. A screenshot of one page's settings covers only that page.

LinkedIn posts from "What happened when we stopped gating everything" (fictional)

Inputs used: full transcript (about 650 words, speakers Priya Shah, Marco Ruiz and host Lena Ortiz), author Priya Shah (Head of Growth, Stackwise), audience B2B SaaS marketers, CTA https://stackwise.test/replay. Speakers: Marco Ruiz is internal (Head of Demand Gen at Stackwise) and fine to quote by name, so no guest approval is needed. Lena hosted; post 10 mentions her question but does not quote her.

Happy to. Every post has to trace back to a real moment in the webinar, so I need three things first:

1. Can you paste the full transcript, with speaker names? At least 500 words. The whole talk is best; a Zoom, YouTube caption or transcription tool file works.
2. Whose LinkedIn profile will post, and what's their role? This changes how quotes are attributed: if you post but someone else said it, I'll write "In our webinar, [speaker] said..."

Subject

Halden Clinics: weekly status, week of 26 October 2026

Status

At risk. The October blog post is blocked waiting on Dr. Patel's clinical review, and its date has slipped from Oct 24 to Oct 31 at the earliest. Everything else is on track.

(Deciding lines: "Blocked: October blog post 'Back pain at work', waiting on Dr. Patel's clinical review since Oct 22" and "blog post date slipped from Oct 24 to Oct 31 at the earliest".)

Done this week

I can write Acme's update, but I don't have enough to write it honestly yet. "A bunch of SEO stuff and some ads" can't become outcomes, and an annoyed client is exactly the one who will check the dates.

What I need (required): 1. This week's tasks with status. Each task marked done, in progress (with its expected date) or blocked (with the reason). Only tasks marked done will appear under "Done this week". 2. Next week's plan. The tasks planned for next week.

Headline numbers

Tidewell Swimwear, Meta. This week 2026-10-19 to 10-25 vs last week 2026-10-12 to 10-18. Conversion: purchase, 7 day click 1 day view, same both weeks. Target CPA $35.00.

Rule: totals are sums of M1 to M3; ratios are recomputed from the sums; changes use the rounded values shown, half up. No total frequency: reach is not summed across campaigns, because one person can be in several.

Campaign table

What moved

I can write the client note, but I need four things first, because the numbers you gave can't be recomputed and one platform's conversions changed definition.

Required:

Welcome sequence: PrepPal meal prep planner downloaders (fictional)

Happy to write it. Each email in the sequence pushes toward one action, so I need three things first:

1. Who signs up, and why? What do new subscribers expect, and where do they sign up (a lead magnet, your site, a social link)?
2. The one action you want new subscribers to take. Which action best predicts they stick around or buy? For example reading a first issue, replying, joining a community or buying something.

Segment definition

• Enters: customers with at least 1 past order whose last order was 150 or more days ago (your lapse rule).
• Needs: a last order date field and an order count field on each profile in Klaviyo or Mailchimp.
• From name: Kettle & Crumb (default: brand name).

Sequence overview

Timing is taken from your input. Day 0 is the day a customer enters the segment.

Email drafts

Happy to write this for Kettle & Crumb. Four required items are missing or undecided, and each one changes the copy:

1. Lapse rule. "A while" isn't a rule your email platform can use. Who counts as lapsed? For example: at least 1 past order and last order 120 or more days ago.
2. Offer. "Maybe 15% or free shipping" is still open. Pick one, with its terms: code, minimum order and end date, or "no discount". I write offer terms exactly as approved, and I only say "ends soon" when there's a real end date.

Settings checked

Site: https://brightpathtutoring.example (WordPress 6.6, Yoast SEO, one author). You did not give the Yoast version, so menu names below may vary by version.

Rows checked: 8. Search result: 3 Match, 5 Conflict. Sitemap result (show in search and in sitemap disagree): 3 Mismatch (S3, S4, S8). Reading setting: "Discourage search engines from indexing this site" is CHECKED.

Findings

I can audit greenleafgoods.example, but three required items are missing or unclear, and each changes the findings.

1. The "Discourage search engines" box (required). In Settings > Reading, is "Discourage search engines from indexing this site" ticked? Since WordPress 5.3 this box adds a noindex, nofollow robots tag, so it matters when a whole site is missing from Google.

URL table

Sitemap: https://www.northfieldoutdoor.example/sitemap-products.xml. Crawl: Screaming Frog list mode, 2026-10-27. Class order applied: Error, Redirect, Noindex, Non canonical, Keep.

Counts by class

Rule: total = Keep + Error + Redirect + Noindex + Non canonical.

6 + 2 + 2 + 1 + 1 = 12, which equals the 12 URL rows you sent.

Remove or replace list

I can audit this sitemap, but two required items are missing, and both change the result.

YouTube metadata audit: Lathe and Grain

Scope: the 8 videos and 4 playlists you pasted. Descriptions were given as their first 3 lines only, so the 5,000 character description limit could not be checked; I counted titles only. Character counts include spaces and were computed by script.

Limits used (Platform facts):

Video checks

I can audit your watercolor channel's titles, descriptions and playlists and propose fixes. Your message did not include the channel data, so I need two things first:

Questions found

Video: 'Setting up Shiftly for a restaurant team', 6:40, https://youtube.example/watch?v=shiftly01. Each answer is quoted from the transcript line at that timestamp.

Questions found: 7, counted from the table.

Buyer question coverage

Counted from the table: 4 buyer questions, 3 answered, 1 not answered.

Chapters

``` 00:00 Setting up Shiftly for a restaurant team 00:22 Import your staff from a CSV 01:05 Do staff need an app? 01:48 Build the week

I can write this description, but not without the transcript. A description that answers buyer questions has to use what the video actually says, and I do not have it.

What I need (required):