Weal Google Ads — the campaigns, v3.4 (2026-10-06; v3.3 2026-10-05, v3.2 2026-10-03, v3 2026-09-23)

Every keyword, blocked search, ad and setting, exactly as it would be built: docs/06-launch/ADS-BUILD-SHEET.md (DOC 6 on the dashboard, pinned at the top), generated from the same file the API calls use. On 2026-10-06 the whole build was sent to Google as a validate-only request — Search 430 operations, Performance Max 37: both valid, nothing created (re-checked the same day after the code review: Performance Max now targets people in the US only, like Search, and keeps out the six coupon words as Search's non-brand groups do).

The plan in one minute

Two campaigns, $120 a day, about $3,650 a month (owner, 2026-10-06: "80 a day on search and 40 on pmax"), inside the $5,000-a-month ceiling — reviewed by four independent AI reviewers on 2026-10-05 (v3.3) and settled by the owner on 2026-10-06 (v3.4; the review table below says what was taken).

Before the first dollar: the test order goes through and shows up in Google Ads as one purchase at the right value, with enhanced conversions on; a second payment method on the ads account; and the new-advertiser credit claimed (the owner found a $500-for-$500 offer under Billing → Promotions, 2026-10-06). Then Claude creates both campaigns paused, the owner looks them over in Google Ads, and switches them on.

The first two weeks: each morning Claude cleans the search terms (new negatives) and checks that Google's purchase count matches Shopify's orders. Week one runs on exact match and "Maximize clicks" with a $3 cap — cheap, controlled traffic while the tracking proves itself; nothing else changes. Day 7: Search switches to "Maximize conversions" and phrase match comes on. Day 8: AI Max's search matching comes on (owner, 2026-10-06: "switch from week 2") — Google may then show the ads on searches beyond our keywords, still inside the blocked list; it never writes the ad text and never sends people anywhere but the product page. Day 14: each campaign is judged on what one sale costs, against what people actually bought: the page pre-selects Buy 2 Get 1 Free on subscription ($79.99, about $54 of contribution), so a sale breaks even at about $50 if buyers keep the default and about $24 if they switch to a single pouch. Under break-even keeps going and grows 20% a week (toward $165/day, the $5,000 ceiling) with the test groups D and E added; over it holds the budget while the page or the keywords are fixed.

The owner's calls (2026-10-06): $80 + $40 a day; the ads keep v2's text word for word — no line changes without the owner's yes on the exact sentence; the money rules raise alerts, they never pause anything; no extra ad features (v2's sitelinks and callouts as they were). Later the same day: the ad text review's changes the owner took are in (DOC 7, D-132/D-133), and AI Max's search matching comes on at day 8.

What the review changed (v3.3, 2026-10-05)

Four independent AI reviewers (kimi-k3, deepseek-v4-pro, minimax-m3, nemotron-3-ultra) read v3.2 in full, each told the settled decisions (two campaigns, $100/day, the product page as the only destination, purchases as the goal, no "brain fog" keywords, AI Max and auto-created assets off, no "Alpha-GPC" in the feed title). Their points were checked against the store before anything was taken: a claim that the lab report does not exist was wrong (it does, /pages/lab-results), and the plan's own assumption that most first orders would be a single subscribed pouch at $39.99 was wrong too: the page pre-selects Buy 2 Get 1 Free on subscription ($79.99), checked live.

Not taken, and why: a separate brand campaign now (2 for, 2 against — brand searches read 0; split it out when Meta starts and people look the name up); "new customers only" in Performance Max (every customer is new this month); a product video (it pushes Performance Max onto YouTube); remarketing lists (month two); stars on the Shopping ads from the reviews (you decided on 22 September to keep stars out of Google); the single-ingredient heads ("l theanine supplement", "lion's mane supplement"), "brain fog supplement" and a "study supplement" family (month-two tests, once the account has history and the volumes are checked). Your call, recommended after the first sales: two reviewers would rewrite the Shopping titles to "Weal Focus Daily Focus Supplement – 60 Capsules (30-Day Supply) – Lion's Mane, L-Theanine, Zinc, B12" (still no "Alpha-GPC"). Merchant Center approved the current titles two weeks ago, and a title change can send the listing back for review, so not before launch.

Every setting, headline, description and the negative list not listed in this file are unchanged from v2 and live in docs/06-launch/SEARCH-CAMPAIGNS-v2-2026-09-22.md.

What changed since v2

Where the numbers come from

August (exact): Google Keyword Planner, US, 12 months Jul 2025 – Jun 2026, pulled 2026-08-08 to 08-10 from the MyBirdWall Google Ads account, which has spend history and so gets exact numbers (docs/04-data/MASTER-KEYWORD-UNIVERSE.csv, 10,478 terms; method in docs/02-screening/DEMAND-FINDINGS.md).

October (ranges): the same 183-term list re-run by the owner on 2026-10-02 in the Weal account, US, all languages, 12 months Sep 2025 – Aug 2026 (docs/04-data/kp-exports/round9-refresh-2026-10-02-weal-account-ranged.csv; the side-by-side is docs/04-data/kp-refresh-2026-10-02-comparison.csv). Weal's account has not spent yet, so Google gives ranges (10–100, 100–1K, 1K–10K, 10K–100K…) and no month-by-month figures.

What the refresh shows: of the 134 terms measured both times, 130 sit in the same range as in August; the 4 that moved were each within about 10% of a range edge (e.g. "lions mane mushroom" 90,500 → 100K–1M). The three check terms land where they did (nattokinase and berberine 100K–1M, DIM 10K–100K). So demand for these searches has not shifted at the precision available, and the August exact numbers stay the working estimates. Bids moved more than volume: alpha-GPC's are unchanged, choline's and "vitamins for focus"'s top ends are up 20–40%. 21 of the 183 came back merged into a close variant (e.g. "alpha-gpc" into "alpha gpc", "lion's mane mushroom" into "lions mane mushroom"). For exact October numbers run the same file in the MyBirdWall account — owner, 2026-10-03: MyBirdWall has no active campaigns either, so it would be approximate too. Re-run the same list (dashboard/downloads/keyword-refresh-2026-10.csv) in the Weal account about two weeks after the campaigns start spending, when Google shows that account real figures; then update these tables.

The keywords, by ad group

Match types: exact only in week one, exact + phrase from day 7 (v3.3). Every ad group lands on the product page. Vol = searches/month.

A · Alpha-GPC — the core (week 1)

B · Choline (week 1)

The biggest buying-intent pool in the list. It was one line inside group A in v2; it gets its own ad group so its own headlines ("The Choline Form: Alpha-GPC") and its own budget share. ⚠️ Its top-of-page bids rose (high end $5.30 → $7.50): under week one's $3 cap these ads will sit low on the page, so the group is judged after day 7, on Maximize conversions.

C · Focus formula (week 1)

D · Brain and nootropic — test tier (week 3, only if A–C buy under break-even on 15+ purchases)

Big and broad: these people have not chosen an ingredient, so they convert lower. Held until the core groups have shown what a purchase costs. "nootropics supplement" is now far more contested (competition 42 → 93 of 100), so it goes last.

E · The pairing — test tier (week 3)

Brand searches (an ad group inside Campaign 1 since v3.2; a separate $3/day campaign before)

weal focus · weal supplements · weal focus supplement · weal focus alpha gpc · weal focus reviews. All five read 0 searches in October — nobody searches the name yet. The ad group costs nothing until they do, and it is there for the day Meta and the Search ads make people look the brand up.

Looked at and left out — and why

Negatives added in v3.3 (phrase match)

• The gym use of alpha-GPC: pre workout, preworkout, bodybuilding, gym, workout, muscle, testosterone, hgh, growth hormone, power output, athletic performance, weight loss, fat loss
• Other ingredients and products: bacopa, ginkgo, ginseng, ashwagandha, rhodiola, tyrosine, dmae, creatine, caffeine pills, energy drink, methylene blue, cbd, melatonin, magnesium, multivitamin, omega 3, tablet, tablets, chewable, softgel, softgels, spray, patch
• Support and after-sale: cancel, cancellation, refund, return policy, customer service, login, order status, tracking, phone number, complaint, scam, lawsuit
• Research and reference sites: forum, forums, quora, examine, healthline, webmd, erowid, longecity, ncbi, nih, clinical trial, systematic review, meta analysis, placebo, withdrawal, tolerance, does it work, how to use, how to take, when to take, best time to take, natural sources, foods, deficiency, dopamine, drug test, banned, fda, legal, safe, safety
• More prescription words: concerta, armodafinil, provigil, nuvigil, donepezil, aricept, memantine, namenda, smart drug, study drug
• Trade and deal hunters: private label, manufacturer, supplier, distributor, dropship, deals, black friday — and, on ad groups A–E only (not the brand searches), discount, promo, promo code
• Shops and rival brands: rite aid, temu, gorilla mind, neuro peak, cognium, hunter focus, noocube

Checked against every keyword above so none blocks one of ours: "sale" stays off the list ("alpha gpc for sale" is a buyer), so do "stack" ("nootropic stack"), "price" and "review".

Budget: $120 a day — Search $80, Performance Max $40 (owner, 2026-10-06)

Google sets a budget per campaign, not per ad group: inside Search, the bids and the pause rule steer the money, and the ad groups' shares are whatever the auctions give them.

Bidding. Search: Maximize clicks with a $3.00 cap and exact match for 7 days → Maximize conversions on day 7, phrase match on → Target CPA at 15+ purchases in 30 days, set at the cost actually seen, then stepped down 10–15% a week toward break-even. Under a $3 cap the choline and "vitamins for focus" terms will rarely show in week one; they are judged after the switch, not before. Performance Max: Maximize conversion value with no target, provided the test order shows Google the real order value (if it shows $0 or a flat value, switch it to Maximize conversions and fix the tag); a ROAS target once it has 15–30 sales.

Performance Max settings (v2's, except): no text, image or video assets — the product feed only (if Google insists on assets: the minimum — business name, logo, a few of v2's headlines; still no video), with asset optimisation all off; final URL expansion OFF; Weal, Weal Focus and Weal Supplements as brand exclusions; the shared negatives, which reach its Search and Shopping placements; its own tracking tag (utm_campaign=pmax); search themes = the A–C keywords; Google's in-market "Vitamins & Supplements" as the one audience signal; account-level exclusions for mobile apps, games and parked domains. Day 10: if more than a third of its money went outside Shopping with almost none of its sales, tighten it. Day 21: a sale costing more than twice Search's, or over $60 with fewer than 5 sales, raises an alert with a proposal (pause it, move its $40 to Search); the owner decides.

Rules for days 1–30 — alerts, never automatic pauses (owner, 2026-10-06). Claude posts each one in #karst and brings it to the owner with a proposal; nothing stops on its own. - Week one: search terms and the Google-vs-Shopify purchase count every morning; no other change. - Alert: a keyword at $60 spent with no purchase and no add-to-cart, or $100 with no purchase; an ad group at $250 with no purchase and add-to-cart under 2% of clicks. - Growth is proposed only on 15+ purchases under break-even; at day 14 a move of at most $10–15 a day between the campaigns is proposed, toward the cheaper one on 8+ purchases. - Reading a weak result: few clicks per impression (CTR under 2%) is the ad; add-to-cart under ~4% of clicks is the page; add-to-carts without purchases is the checkout, the price or the shipping. - Money never moves to Performance Max on its own reported cost alone — it moves when total orders rise. - A disapproval: the exact replacement line goes to the owner first (owner, 2026-10-06), never a change to the page's claim; if Google questions the product page itself, the appeal goes in the same day with /pages/citations as the support. - Break-even is read from the orders that actually came in: about $50 a sale if buyers keep the pre-selected Buy 2 Get 1 Free subscription, about $24 if they pick a single pouch.

Ads and assets — v2's, word for word (owner, 2026-10-06)

The three responsive search ads, the four sitelinks and the five callouts are v2's, unchanged; the build sheet (DOC 6) lists every line. Two things the API dry run turned up:

• One brand description was 92 characters; Google allows 90. Proposed: "300 mg Alpha-GPC, lion's mane, L-theanine, zinc, B12. Every amount on the label." (81) — the owner's line (2026-10-06, D-131): "Weal Focus: 300 mg Alpha-GPC, lion's mane, L-theanine, zinc, B12." (65 — v2's first sentence; its second, "Every amount on the label", is already one of the brand ad's headlines). Google re-checked the build with it: valid.
• The business name and logo wait for Google's advertiser verification — linking them on an unverified account is refused. Everything else builds without it. The owner left verification for later (2026-10-06).

History: v3.3 proposed taking three lines out and adding a price line, more callouts, a snippet, a promotion and new sitelinks; the owner declined all of it on 2026-10-06.

Before the first dollar

• [ ] The test order goes through and appears in Google Ads as one purchase, at the amount charged, with enhanced conversions active (Google Ads → Goals → Conversions → diagnostics); GA4's purchase stays secondary
• [x] Auto-tagging on (read through the API 2026-10-06)
• [ ] The product URL with ?variant= plus the tracking suffix opens correctly
• [x] A second payment method on the ads account (the owner, 2026-10-06); the $500-for-$500 promotion is there (the owner found it on 2026-10-06; at $120 a day the $500 is spent in about four days)
• [ ] Google's advertiser verification started — later, the owner's call (2026-10-06): it unlocks only the business name and logo, the owner did not find it under Billing, and Google emails a deadline when it wants it; the API cannot read its status on card billing (it answers only for monthly invoicing)
• [ ] Settings read back in the Google Ads screens: Search only, partners and Display off, "Presence", AI Max off at launch (all three parts; search matching on at day 8, the other two parts never), automatically created assets off, the negatives attached and none blocking a keyword
• [x] Auto-apply recommendations: none switched on — the account has no auto-apply subscriptions (read through the API 2026-10-06); re-read once the campaigns exist
• [ ] Performance Max read back: feed only, asset automation off, final URL expansion off, "Presence", the brand and coupon searches blocked, its own utm_campaign=pmax tag, $40/day
• [x] Built paused through the API on 2026-10-06 at the owner's word ("build the campaigns paused"); the read-back matched every item; the ads are in Google's review — switching on is the owner's click
• [x] Enhanced conversions on in Shopify's Google & YouTube channel (the owner, 2026-10-06)

Weal Page Review — Buyer's Eye (2026-09-22)

Repo copy of the Claude Doc https://claude.ai/code/artifact/2c57070e-32ae-482d-b60b-08e9b9727ab1 (Claude Docs belong to the Claude account that made them, so this file is the durable copy). Three reviewers walked the live store on a phone: a 54-year-old from the Facebook ad, a 34-year-old comparing from Google, and a claims-and-payments auditor. Full reports: docs/06-launch/reviews-2026-09-22/.

Owner's answers (2026-09-22, D-119)

• Subscription policy: ✅ written and live (set_policies.py, SUBSCRIPTION_POLICY).
• Slide 1 "FREE SHIPPING & RETURNS" with paid return postage: keep.
• hCaptcha badge over the sticky Add to cart: keep for now. Superseded 2026-10-03 (D-126): the floating badge is hidden and its notice sits under the popup's form; the protection stays on.
• Lab page heavy-metal sentence: keep.
• Buy-box renewal wording: keep.
• Alpha-GPC is the 99% grade: ✅ "99% pure" on the description, ingredient card and lab page.
• Lot 20260701: ✅ labelled "(the manufacturer's batch code)" on the lab page.
• V2 live for Google's review: no rating row, counters, timeline or reviews. Google approved the products; V1 is live again since 2026-09-26 (D-121), so the timeline lines C6–C7 are on the page again.

Owner's answers, round 2 (2026-10-02, D-123) — all applied and checked on the live site

The slideshow dots stay 18px: six dots in a row cannot each get a 44px area without spreading them 2.5× wider across the picture, and the thumbnails and a tap on either side of the picture already change it.

Copy still open — yes / no / edit by number (2026-09-22 list, kept as the record)

The owner's answers on the N items (2026-10-03) — settled

The N items, explained (2026-10-02)

Each one: what it is, why it matters, what the fix is, and who does it. Two are already done.

N1 — The FDA sentence, visible without a tap. Your yes. What: "These statements have not been evaluated by the Food and Drug Administration…" sits inside the product page's "Important" section, which starts closed — a visitor only sees it if they tap it open. Why: a page that says what a supplement supports is expected to carry that sentence where the claims are, and Google's and payment reviewers look for it; behind a tap is the weakest place for it. Fix: show the same sentence once, in small print under the three badges, without a tap. No new words. The Cognitive Review stays without it (your standing decision). Recommendation: yes.

N2 — Garbled small print on the pouch in slides 2 and 3. Your yes. What: the pouch in those two pictures was generated, and its tiny print came out wrong — zoomed in it reads "Vitamin B13", "80 CARSULES", "S0 SERVINGS". Slide 1 had the same problem and was fixed. Why: anyone who zooms sees a count and a vitamin that are not on the real pouch; it reads as fake, and "not as pictured" is a chargeback reason. Fix: paste the real label strip from the pouch artwork over that area in both pictures. Recommendation: yes.

N3 — Slide 2 says more than the page. Your yes. What: slide 2 reads "Over 100%+ Improvement on Attention Tests". The claim you kept on the page says "…more than double attention test gains versus the next best option"; the slide drops that last part, so it reads as "doubles your attention". Fix: add the line "vs. the next best option" under "Attention Tests" — your own kept words, nothing new. Recommendation: yes.

N4 — Small overlaps on the slides on a phone. Your yes. What: on a phone the dots sit on slide 2's study line, and the arrows touch letters on slides 3 and 6. Fix: move those words inside the pictures a little away from the dots and arrows. No words change.

N5 — 99% alpha-GPC. ✅ Done 2026-09-22.

N6 — What the next lab test should add. You, with the next batch. What: the current report checked 5 metals and 6 microbes, all clear. A careful buyer could still point at three gaps: the microbe tests are the cosmetics set and include no Salmonella test; nothing measures how much alpha-GPC is actually in a capsule (only the blend's proportions); the lead and cadmium limits are coarse. Fix: when you order the next lot, ask Max's lab for the supplement microbe panel with Salmonella, a potency test of the actives, metals at lower limits, and the report as a PDF. Nothing to do now.

N7 — The paperwork behind "Made in a cGMP facility". You — one message to Max. What: the site says it in three places; we hold no certificate for it. Why: if Google, a payment processor or a customer asks, you need the document. Fix: ask Max for the factory's cGMP certificate or registration and keep it on file (not published).

N8 — The lot number. ✅ Done 2026-09-22 (labelled the manufacturer's batch code).

N9 — The checkout looks like a different store. You (10 minutes), or say so and I set it. What: Shopify's checkout shows the store name as plain text, Shopify's blue buttons and a white page; the "Email me with news and offers" box is ticked before the buyer touches it. Why: payment is where trust is lowest; a pre-ticked marketing box annoys careful buyers. Fix: Settings → Checkout → Customize: logo, buttons in the brand green, cream background; and untick "pre-select" under the email marketing option.

N10 — "Pay in 4" on a subscription. Your yes. What: the cart says "4 interest-free installments…" next to a subscription; Shop Pay can't split a subscription, so it advertises something the buyer can't do. Fix: hide that line in the cart. Recommendation: yes.

N11 — The "free" bundles cost a cent more. Your call. What: Buy 2 Get 1 Free is $89.99 but two single pouches are $89.98; Buy 3 Get 2 Free is $134.99 against $134.97 (subscription: $79.99 vs $79.98, $119.99 vs $119.97). Why: a sharp shopper can say the "free" pouch isn't free; the FTC's rule on "free" offers says the paid items must not cost more than their usual price. Fix: price the bundles at exact multiples ($89.98 / $134.97; $79.98 / $119.97 on subscription); the Google feed follows on its own. Recommendation: yes — two cents.

N12 — The chat bubble. You, 5 minutes. What: a Shopify Inbox chat bubble appears after about 30 seconds. If nobody answers, the buyer who asks "is it the 99% form?" waits and leaves. Fix: in the Shopify Inbox app set an automatic reply ("We reply by email within one business day — hello@wealsupplements.com") and keep the app on your phone, or turn the chat off until someone can answer.

N13 — The ranking's table on a phone. Your yes. What: on the alpha-GPC ranking (the Facebook ad page) the comparison table shows 2½ of its 5 columns on a phone; the price column is off-screen and nothing says to swipe. Fix: show each product as a small stacked card on phones (as the dashboard now does), or move the price column next to the name. Recommendation: the stacked cards.

Phone polish

Sticky bar covers the last 83 px · popup targets under 44 px · small tap targets · Buy 1 per-month line · Citations page 12.8 px · ranking small labels · review dates format.

Weal SEO pass — what shipped, what is left (2026-09-22 → 2026-10-02)

Repo copy of the "Weal SEO Pass" Claude Doc (https://claude.ai/code/artifact/01b30919-d2b0-4f3a-9475-7f7b246934c5), which belongs to the Claude account that made it. Decisions: D-118, D-120, D-121. The live list of titles and descriptions is content/seo/seo-fields.json, written to Shopify by scripts/shopify/set_seo_fields.py.

Shipped and live

Left (none of it urgent)

• Bing Webmaster Tools — later (owner). One-click import from Search Console when wanted.
• Google Ads ↔ Search Console association — optional; Search Console → Settings → Associations → Google Ads 946-201-9556.
• The empty "News" blog — still live at /blogs/news and in the sitemap with no posts; delete it in Shopify admin → Blog posts → Manage blogs → News (owner's click).
• Research-index thumbnails keep alt="" on purpose: each sits beside a link with the same title, so an alt would be read twice.

Looked at and cut (not to be re-proposed)

An AI-crawler robots.txt (a named group replaces Shopify's defaults and opens the checkout paths) · a /pages/llms page · FAQ markup (Google retired FAQ rich results on 7 May 2026) · a star rating in the product's structured data.

Every title and description

Store (7)

Journal (20)

The Cognitive Review (24)

Paid ads plan — where it lives and what it asks

Current status (2026-10-03). The plan below is the research from 2026-09-22; the build that follows from it is docs/06-launch/SEARCH-CAMPAIGNS.md v3.2: $100/day — Search $50 + Performance Max $50 — inside the owner's $5,000/month ceiling (D-126). Merchant Center cleared the prescription-drug flag by 2026-09-23 (D-120; re-read 2026-10-03). No campaign exists yet (read 2026-10-03); the test order is the gate. The Claude Docs linked below belong to another Claude account and cannot be read from this box (2026-09-23) — the repo files are the copies, and the dashboard's START HERE block shows them.

The plan was a Claude Doc (unreadable from here since 2026-09-23; this file carries its substance): https://claude.ai/code/artifact/176a372a-bcad-43d6-80f4-184452f0e791 — "Weal Paid Ads Plan", 2026-09-22.

Written on the owner's brief of 2026-09-22 (research and design the first paid-advertising strategy, challenge the Performance-Max-first idea a friend at Hill Health described). Research, inspection and planning only: nothing launched, no account changed, no page edited.

The recommendation, in five lines

1. Google Search first, not PMax — three keyword-controlled campaigns (alpha-GPC buyers → PDP, comparison intent → the ranking only with a publisher line, brand) at ~$50/day for 28 days v3: one Search campaign, five ad groups, ramped from ~$75/day, plus a $3/day brand campaign v3.2 (owner, 2026-10-03): one Search campaign at $50/day — ad groups A–C plus the brand searches, every one landing on the product page (no publisher line, owner 2026-09-22); the D and E test groups join in week 3.
2. PMax and Shopping wait until Merchant Center clears the listing (currently flagged "Healthcare and medicine: Prescription drugs", all three products NOT ELIGIBLE) and the account has ~30 purchases/month — Merchant Center cleared (2026-09-23). v3 starts PMax in week 3 at ~$50/day once 5+ purchases are recorded v3.2: Performance Max runs from day one at $50/day (owner, 2026-10-03), judged after two to three weeks since it starts with no sales history. Google's own PMax guidance (3× CPA/day for six weeks) is ~$7,000 here.
3. Meta after the search month, sized by the conversion rate that month measures (D-109 stands).
4. Economics (D-114): break-even cost per purchase ≈ $32 on a single pouch, ≈ $55 on a realistic mix (one-time orders); break-even ROAS ≈ 1.4; at ~$2/click the funnel needs ~4% CVR on a mixed basket. The campaigns doc reads it on subscription first orders after refunds — ≈ $34 on a cold-search mix — and that is the figure v3.2 judges each campaign against at day 14 (keep under $46).
5. Six page fixes before the first dollar — one-time price printed and defaulted for paid traffic, the "more than double" claim reworded, FDA line + retitle on the ranking, popup off on paid landings, zinc/B12/absorption/liver/caffeine lines corrected, SEO title set. Owner, 2026-09-22: done — the one-time price printed (option B), the liver line, the product title, the SEO titles; kept as they are — the claim, the ranking, the popup and the other lines (D-115).

What the research established (detail and citations in the doc)

• The friend's "search terms route searches inside PMax" is wrong on Google's own docs: search themes are "optional and additive", phrase-match priority, no routing guarantee; only separate campaigns fenced with negatives (or Search exact keywords, which beat PMax) route reliably.
• Search themes/audience signals guide, never restrict. Final URL expansion off is the only landing page guarantee, and feed product ads still land on the feed's page.
• Health is a Google sensitive interest category: no advertiser-built audiences; no copy implying a condition.
• Claims: "more than double the next best option" is unsupported as worded (the comparator was placebo; Kerksick 2024, n=20, one dose); "most readily absorbed" is contradicted (Böckmann 2022); the caffeine pairing belongs to L-theanine, not alpha-GPC; "moves fat out of your liver" reads as a disease claim; the ranking carries no FDA disclaimer. Lee 2021 (stroke cohort) sits in the file.
• The undisclosed publisher on the ranking is the FTC review rule's §465.6 case when used as an ad destination; the 688 reviews / counters are the Merchant Center exposure. Both are the owner's standing decisions and are listed as exposure, not re-argued.

Status, 2026-09-22 night

Done since the plan: product title "Weal Focus"; the owner's description (D-116); one-time price printed (option B); category + MPN set in Merchant Center; appeal submitted on all three offers (prescription drugs, "I disagree"). Kept by the owner: the H1, the buy-box claim, the popup, the ranking as it is. Next big item: the SEO pass. ✅ Shipped the same night (D-118); V2 of the product page went live for Google's review (D-119).

The two companion docs (2026-09-22, evening)

History: both are superseded in their working form — the campaigns are v3.2 in SEARCH-CAMPAIGNS.md (2026-10-03) and the SEO pass shipped the same night (SEO-PASS.md, D-118). Both Claude Docs are unreadable from this box since the 2026-09-23 account swap.

• Weal Search Campaigns — https://claude.ai/code/artifact/0b03d311-4983-45be-8405-7e6aa4600a9e — the build in full: one Search campaign (three ad groups: alpha-GPC incl. the head term, focus formula, choosing) at $47/day + a $3/day brand campaign; a retail PMax built and paused with no date (needs Merchant Center approval, 30 purchases/30 days and ~$2,520 of new money); every keyword, negative, headline, description, sitelink; the economics read on subscription first orders (break-even ≈ $24–48 after refunds, ≈ $34 on a cold-search mix; Google never sees renewals); AI Max and automatically-created assets OFF. Reviewed twice by three council reviewers (Google Ads practitioner, supplement-policy reader, performance analyst): all three agree to create as written. Creation waits on the owner's go and the test order.
• Weal SEO Pass — https://claude.ai/code/artifact/01b30919-d2b0-4f3a-9475-7f7b246934c5 — 56-page crawl baseline, a 10-step ship order (Search Console + Bing first, money-page titles, the Review's share image, the doubled description tag, the 688-card product page, sitemap hygiene, structured data, then the 42 editorial descriptions + 20 search titles from a repo file), the AI-search section with the playbooks' evidence read at source, and the items cut after review (AI-crawler robots, llms page, FAQ markup, star-rating markup). Owner decisions: the alpha-GPC ranking's search title (keep "Healthy Aging" or not), the 10-card reviews section, the star rating (recommended no), Search Console/Bing accounts.

Owner decisions the plan waits on (BLOCKERS 🟡)

Budget ceiling ($5,000/month for Google, owner 2026-09-23 "for now"; $100/day to start, 2026-10-03, D-126 — the PO#1 question no longer gates anything once the budget is monthly) · publisher line on the ranking (no — the ranking stays as it is, owner 2026-09-22; campaign 2 therefore lands on the product page) · default offer and price display (option B, done) · Merchant Center review request (submitted, and the flag cleared by 2026-09-23) · Meta in parallel or after · the test order · the owner's yes on v3.2 (DOC 1) (settled: v3.4, 2026-10-06 — the exact build is DOC 6), then the go to create both campaigns paused.

Repo facts used

Keyword Planner exports docs/04-data/kp-exports/ (Aug 2026); unit economics scripts/shopify/catalog.py (D-114); conversion actions, Merchant Center statuses and account list read by the Google Ads API (~/.gcp-creds/pgovteam-ads-token.json, D-113).

Taxes — sales tax first, then income tax and the rest (2026-10-03)

Research for the owner and an accountant, not legal or tax advice. Every rule below links to its source; the store facts were read from Shopify the same day.

Where the store stands today

• Shopify charges $0 sales tax everywhere: no state registration is entered, so it collects nowhere (a checkout calculated for New York, California and Ohio on 2026-10-03 shows $0.00 tax).
• The business address and the Shopify location are in Cleveland, Ohio.
• The stock sits in Max Dropshipping's US warehouse. Its Shopify location carries no address, so neither we nor Shopify can tell which state it is in.
• The product has no Shopify product category (type "Dietary supplement"). Shopify uses the category to apply supplement rules; uncategorized products get the general rate.
• The store was created 2026-08-11, which decides the Shopify Tax fee tier (below).

The two ways a state can make you collect

1. Being there (physical presence) — from the first sale, no threshold.

• Ohio, because the business is there. Every Ohio seller needs a vendor's license, online sellers included, before making taxable sales; selling without one is an offense, and a return is due every period even with no sales (Ohio Dept. of Taxation, brochure). The license costs $50 since 2025-04-09 (ODT). Only orders delivered in Ohio are taxed (ST 2009-03).
• The warehouse's state, because the stock is there. In nearly every sales-tax state, inventory in a warehouse counts as being present — California, for example, treats an out-of-state seller with stock in a California fulfilment center as doing business there (CDTFA). The exceptions mostly cover stock the seller does not control or marketplace-only programs, which a 3PL shipping our own orders is unlikely to fit (Avalara, updated 2026-05-01). Delaware, Montana, New Hampshire and Oregon have no sales tax.

2. Selling enough into a state (economic nexus) — only above a threshold.

Most states: $100,000 of sales a year into that state. The ones that differ (Streamlined Sales Tax, cross-checked with TaxCloud, 2026-09-22):

Several states dropped the 200-sales test recently — Illinois on 2026-01-01 (IDOR) and Kentucky on 2026-08-01 (HB 757) among them.

At our size this is far off: at $100 a day of ads, a few thousand dollars a month of sales is roughly 600 orders a year across every state, so no single state comes near $100,000 or 200 sales. Shopify watches it for us (below).

Is a supplement taxable?

It depends on the state. Weal Focus is a "dietary supplement" by the national definition (the Supplement Facts panel), and each state decides whether that counts as tax-free food (SSUTA).

How Shopify fits in

• It collects only where you tell it to. Settings → Taxes and duties → United States → Add state → enter that state's sales tax ID → collect (Shopify Help).
• It watches the thresholds. The same page has "Tax liability insights": it flags states where you are present (from your location addresses — so Ohio shows, the warehouse's state does not until Max's location has an address) and puts a state on "Monitoring" at 80% of its sales threshold (Help).
• It does not file or pay. Returns and payments go to each state from you or your accountant (Help).
• The fee: stores created after 2026-05-13 (ours: 2026-08-11) get their first $100,000 of sales free of the Shopify Tax fee; after that it is 0.35% of each order in the states where you collect, at most $0.99 an order (Pricing).
• Shipping: we charge none, so there is no shipping charge to tax.

What to do, in order

1. Ohio, before Ohio orders: get the vendor's license through OH|TAX eServices ($50), then add Ohio in Shopify with the license number. Ohio taxes an Ohio seller's in-state orders at the rate where the order is received (ST 2009-03), which may mean Cuyahoga County's 8.00% on every Ohio order — check what Shopify charges with the accountant.
2. Ask Zoe one thing: which state the warehouse is in (and whether stock ever moves between warehouses). If that state taxes supplements, register there before shipping orders into it; if it exempts them, ask the accountant whether you still need to register.
3. Set the product's category to Vitamins & Supplements in Shopify when the first state is added, so the state rules (exempt in Texas, taxable in Ohio) apply. Not before: it is a product edit Merchant Center may look at again, and with no state on it changes nothing today.
4. Everything else waits for "Monitoring" on the liability-insights page.

Income tax and the rest (added 2026-10-03, owner's question)

• Federal income tax. An LLC usually pays none itself: a one-owner LLC's profit goes on the owner's own return (Schedule C); an LLC with several owners files a partnership return and each owner gets a K-1; an LLC that elected S-corp files its own return and pays its owner a salary. Which one pgov LLC is decides the forms — the accountant will know.
• What gets taxed is profit, not sales: sales, minus refunds, product cost (the PO#1 pouches count as they sell, not when they were bought), ad spend, Shopify and card fees, Max's shipping and the rest. Sales tax collected is not income — it is passed on to the state.
• Self-employment tax (Social Security and Medicare, about 15.3%) applies on top of income tax to the profit of a one-owner or partnership LLC.
• Quarterly estimated payments are due once you expect to owe $1,000 or more for the year (mid-April, mid-June, mid-September, mid-January).
• 1099s. Shopify Payments sends a 1099-K only above $20,000 and 200 sales in a year — the 2025 law put that limit back (RSM); every sale is income either way. You send a 1099-NEC to a US freelancer you pay $2,000 or more in 2026 (it was $600; OnPay) — not for product, ads or Max.
• Ohio income tax (if you live in Ohio): a flat 2.75% in 2026, and the first $250,000 of business income is exempt; above that, 3% (Uncle Kam).
• Cleveland city tax: 2.5% of the business's net profit if it operates in Cleveland, the store's address — filed with the Central Collection Agency (CCA). Ohio's $250,000 exemption does not apply to city tax.
• Ohio's gross-receipts tax (CAT): nothing to pay or file under $6 million a year of Ohio sales (Ohio Dept. of Taxation).
• The federal ownership report (BOI) no longer applies to US-formed companies, since March 2025 (FinCEN).
• Registering the company is a different thing from registering for sales tax (owner's question, 2026-10-05). An LLC formed in Ohio is already registered there, with no yearly report or fee. An LLC formed in another state but run from Ohio registers once as a "foreign LLC" with the Ohio Secretary of State ($99, Form 617, with an Ohio statutory agent — Ohio SOS fee schedule). Check where pgov LLC sits at businesssearch.ohiosos.gov. Selling online into other states does not mean registering the company there; a state's sales-tax permit, if one is ever needed, is separate.
• Import duties are a cost, not a return: Max ships from China (Hangzhou), and duties are paid by whoever imports — ask Max whether PO#1's $259.09 freight included them, and for every re-order.

For the accountant: what kind of LLC pgov is for tax (one owner, several, or S-corp), where you live, and whether the business counts as operating in Cleveland; whether stock arriving in the warehouse state on 2026-09-22 created an obligation from that date (no sales yet, so no tax was due on sales); which Ohio rate applies; how often to file in Ohio; and whether subscription renewals pick up tax once a state is added.

Not verified: the Ohio statutes themselves (the state's code site refused connections, so these rules come from the Department of Taxation's own documents); Max's warehouse state (not published); the list of inventory exceptions beyond Oklahoma's; and three Shopify behaviours — Ohio's sourcing rule on orders shipped from out of state, tax on renewal orders our billing job creates, and Illinois' medicinal-claim test.

Ad text review — every proposed change, line by line (6 October 2026)

You asked for the ads' headlines and descriptions to be peer reviewed in rounds, to find what is best, what is needed and what would work better. Nothing in the build changes until you say yes — answer "yes to all", or "yes except C2 and C6", or rewrite any line yourself.

Your answers (6 October) — all in the build (DOC 6), and Google accepted it:

• Taken: C1 (Alpha-GPC, 99% Pure), C2 (Buy 3 Get 2 Free Bundle), C3 and C7 (From $39.99, Cancel Anytime), C8 (with your morning coffee), C9–C14 (the brand ad filled to 15 headlines and 4 descriptions) and C16 (no "each lot" promise: "Tested by an independent lab for heavy metals and microbes. Full report on the site.").
• Not taken — your lines stay: C4 (Zinc & B12 Included), C5 (Memory & Focus Supplement), C6 (Zinc Supports Normal Cognition), C15 (Alpha-GPC Choline, 300 mg).
• Notes: ad status checked once the campaigns exist; the "Free" bundle cents — no change (your call).

How it was reviewed

• Round 1 — four reviewers, four angles. Four AI models from four different makers (Moonshot, DeepSeek, MiniMax, NVIDIA) each read every line of your three ads with the product page's own text, the prices, the lab report and the rules: one as a Google Ads copywriter, one as a supplement-ad compliance reader, one as the buyer who just searched, one as a competitor strategist. Each gave keep / improve / cut for every line and its own best version of each ad. All four returned full reviews.
• Merge. Your line stayed unless two or more reviewers, or a fact on the product page, said otherwise.
• Round 2 — the same four judged the merged proposal. All four agreed with every change. Three of four asked for two more swaps (C2 and C13), which are in.
• Round 3 — fact and Google-policy check by a model from a fifth maker (Anthropic): every claim checked against the product page and Google's supplement, misrepresentation and editorial rules, and the FTC's. No line fails; all three ads pass with notes. It found one more line worth fixing (C15), one optional change (C16) and three things to watch, below.
• Google's own check. The three proposed ads, C15 and C16 included, were sent to Google as a check-only request: accepted, nothing created.

The changes — 15 lines, and one optional

Most of your lines stay: 10 of 15 headlines and all 4 descriptions in the Alpha-GPC ad, 12 of 15 and 3 of 4 in the Focus ad, 10 of your 11 headlines and all 3 descriptions in the brand ad (which was not full).

Three things to watch (no wording change)

• The word "Alpha-GPC" may trip Google's prescription-drug rule, as it did in Merchant Center before your appeal cleared it (D-115/D-116). Google's check-only request does not run the full ad review. Once the campaigns exist, I read each ad's policy status, and appeal with the same case if one is flagged.
• The "Free" bundles cost a cent or two more than the pouches you pay for. Buy 2 Get 1 Free is $89.99 one-time, against 2 × $44.99 = $89.98; the same 1–3 cents on Buy 3 Get 2 Free ($134.99 against $134.97) and on subscription ($79.99 / $119.99 against $79.98 / $119.97). The FTC's guide to "free" offers expects the paid pouches at their regular price. It is far too small to matter in Google's review; the clean fix is the store prices, not the ad text — your call, nothing changed.
• "Cancel Anytime" has to be literally true. Subscribers cancel from their account page (Shopify's customer accounts). With your test order, open the account page and check the subscription shows a cancel option — do not press it, since that order's renewal is our real-card test.

Looked at and not taken

• Pin only one headline to line 1 (three reviewers in round 1). Google's own guidance is two or three per pinned slot, so any of them can show; one pin would put the same line first in every ad.
• "9 in 10 Adults Low on Choline", "1 in 2 Over 60 Low on Zinc" — "low on" turns the page's "less than the adequate intake" into a deficiency claim, and the zinc line drops your page's "almost".
• "Lot-Tested: Heavy Metals ND" — "ND" means nothing to a buyer.
• "Alpha-GPC, Not Bitartrate", "Choline That Reaches The Brain" — a comparison and a body-function claim in the ad itself; they read better on the page, where the explanation sits next to them.
• Display path changes (/alpha-gpc/300mg, /focus/formula) — small; yours kept.
• Cutting "Weal Supplements" from the brand ad — kept: brand searchers look for the name.
• Shortening lines at exactly 30 or 90 characters — not needed; Google allows exactly 30 and 90 and accepted them.
• "US adults" → "American adults" in the choline line — same meaning; yours kept.
• A longer brand description 1 — you chose it short on 6 October.
• "From $39.99 On Subscription" instead of "From $39.99, Cancel Anytime" — round 3 found the price accurate as written ($39.99 is on the page, struck beside $44.99); this is only if you want the condition named.
• Good lines with one reviewer behind them — "Up To 46% Off One-Pouch Price", "Ships In 2-3 Business Days", "Lion's Mane, Not Mycelium", "B12 As Methylcobalamin, 500 mcg", "Heavy Metals & Microbes Clean". Each would push out a line with more support; any can be swapped in on your word.

The three ads in full, as proposed before your answers (the build in DOC 6 has your answers)

Alpha-GPC (ad groups A, B) — shown as wealsupplements.com/alpha-gpc/focus

Focus formula (ad groups C, D, E) — shown as wealsupplements.com/focus/supplement

Brand — shown as wealsupplements.com/weal-focus