Aetheria-Labs1/storefront-skills / skills/optimize
optimize
FlaggedScore an existing Lexsis page against the same page-type, design, proof, offer, asset and copy rules used to build pages, propose improvements as a strict optimization plan with real replacement copy and asset decisions, wait for approval, then apply the approved changes with version protection and hosted QA.
references/copy/message-match.md
View on GitHub# Message match
The contract between an ad (or a search query) and the page it sends the
visitor to. `/plan-page` fills the scorecard in section 4 for every
ad-driven or search-driven page and writes the one-line "Message match"
summary the plan requires; `/design-page` re-checks the hero against it in
Compose step 8; `/optimize` runs it first on any paid-traffic page with a
high bounce. `references/traffic-source-meta.md` ("The #1 Rule: Message
Match") gives the tone guidance; this file is the checkable version and
wins where they differ. Channel specifics stay in the traffic-source files.
Severity BLOCK, FAIL, WARN. Tag LAW, RESEARCH, OPERATOR, HEURISTIC.
## 1. Evidence
- Unbounce audit of 300 paid ads: 98 percent failed message match (page headline did not match the ad headline). https://unbounce.com/ppc/poor-message-match/
- Campaign Monitor via Unbounce dynamic text replacement: mirroring the searcher's exact verb in the headline lifted trial sign-ups 31.4 percent (77 days, over 100 conversions per variant). https://unbounce.com/landing-page-examples/built-using-unbounce/dynamic-text-a-b-test/
- Moz case: matched ad and page raised conversion rate 212.74 percent and cut CPA 69.39 percent. https://moz.com/blog/message-match-conversion-rates
- Google Ads landing-page experience (a Quality Score component): relevance, transparency, ease of navigation, mobile, speed; "keep messaging consistent from ad to landing page", "mirror the call-to-action". https://support.google.com/google-ads/answer/7636512 ; https://support.google.com/google-ads/answer/6167130
- Meta: strong quality and engagement rankings with a low conversion ranking signal a post-click disconnect; Meta downgrades ads that withhold information or use sensationalised language. https://www.facebook.com/business/help/436113280262012 ; https://www.facebook.com/business/news/reducing-low-quality-ads-on-facebook
- Google: a one-second mobile delay can reduce retail conversions by up to 20 percent. https://support.google.com/google-ads/answer/7543502
## 2. The contract
Every row applies to paid social, paid search and email or SMS clicks with
a specific promise. "Ad" means the creative or the message the visitor
clicked; for search it means the query and the ad text.
| Id | Element | Rule | Sev | Tag |
| --- | --- | --- | --- | --- |
| MM1 | Headline | The h1 carries the ad headline's core noun phrase and the same promise. Either identical, or a rephrase no longer than the ad with at least 60 percent content-word overlap. Never a broader or a bigger promise. | FAIL | RESEARCH |
| MM2 | Offer | Same discount depth, same gift, same bundle, same price, same code. The offer appears above the fold on `mof` and `bof` types; the code is auto-applied through the CTA href (`?discount=CODE`) and named in the microcopy. | BLOCK | LAW, OPERATOR |
| MM3 | Product and variant | The hero shows the SKU and variant (colour, flavour, size, pack) from the creative; the buy box defaults to it; the cart receives it. Sold-out variants are labelled, never swapped (DP11). | BLOCK | LAW |
| MM4 | Angle | The mechanism, claim or hook the ad leads with is the first claim on the page, in the h1 or subhead. A page that switches from the ad's "no aluminium" angle to "smells great" fails. | FAIL | RESEARCH |
| MM5 | Persona | The audience the ad addresses ("new mums", "runners over 40", "oily skin") is named in the h1, subhead or qualifier within the first screen at 390. | FAIL | OPERATOR |
| MM6 | Tone | Register matches: UGC or creator ad gets a UGC-led hero and first-person copy; clinical ad gets a clinical hero; conversational ad gets conversational copy. Record the tone from the supplied campaign material. | WARN | OPERATOR |
| MM7 | CTA verb | The primary CTA's first word equals the ad's CTA verb (Google: mirror the call to action). "Shop" in the ad and "Add" on the page is a mismatch; use the checklist pattern in the ad too, or match the page to it. | FAIL | RESEARCH |
| MM8 | No new claims above the fold | Every claim in the first screen at 390 is either in the ad or a verified ledger row. Nothing the visitor did not click for appears before the fold. | BLOCK | LAW |
| MM9 | Awareness | Cold social ads land on `advertorial`, `listicle`, `ad-landing-page`, `ugc-creator-collab` or `video-sales-page`; retargeting and brand search land on `pdp`, `pdp-hybrid-landing`, `offer-page` or `retargeting-warm`. Never send a most-aware click to an educational page or a cold click to a buy box with no premise. | FAIL | RESEARCH |
| MM10 | Visual continuity | The hero uses the same creative asset or the same product, variant and setting as the ad. Colour matching is done within the brand palette only: the page never adopts an ad's accent that is not in the plan's palette table (design-rules N14 and A1 win over the meta file's "extract from creative" advice). Fonts stay the brand's (N4). | FAIL | RESEARCH |
| MM11 | Speed and transparency | LCP under 2.5 s on 4G at 390; business name, contact, returns and privacy reachable from the page; no hidden fees (DP12). | FAIL | RESEARCH, LAW |
| MM12 | No sensationalism | Neither ad nor page uses "shocking", "miracle", "doctors hate", "you won't believe", stacked punctuation. Meta quality ranking penalises it; CP2 and CP21 catch it on the page. | FAIL | OPERATOR |
| MM13 | Consistency across variants | With A/B or personalisation variants, every variant keeps MM1 to MM8; only the element under test changes. | FAIL | OPERATOR |
## 3. Procedure
1. Ask for the exact ad image or video frame, primary text, headline, CTA, and offer when they are not already supplied in the request.
2. View the supplied material and record its headline, claims, colours, CTA, tone, persona, product, variant, and scene.
3. Reconcile its audience language with the approved `persona.md` entry or merchant-supplied audience description. Keep every claim inside the ledger.
4. Record in `page record` under `campaign`: `adHeadline`, `adOffer`, `adVariantId`, `adAngle`, `adClaims[]`, `adCta`, `persona`, `tone`, `creativeAssetId`.
5. Fill the scorecard (section 4) in `page plan`. Any BLOCK row marked "no" stops the plan until the merchant confirms the ad change or the page change. Write the one-line summary:
`**Message match.** h1 mirrors "<ad headline>" (overlap 0.8); offer 15% first order auto-applied; variant gid://.../456 in hero and buy box; CTA "Get 15% off my first order"; persona "night-shift nurses" in subhead; tone UGC.`
6. `/design-page` re-reads the page record before writing the hero and runs the section 6 scripts after compile.
7. For an ad claim that cannot be substantiated from the ledger (MM8): the page does not repeat it, the plan flags it under the Claim gate, and the merchant is told the ad itself needs the evidence or a rewrite.
## 4. Scorecard
Copy into `page plan` under `## Message match`. One row per creative element; "Match" is yes or no; a "no" on a BLOCK row stops the plan.
```markdown
## Message match
| Creative element | Value in ad | Page element | Value on page | Match | Rule |
|---|---|---|---|---|---|
| Headline | "Finally, a deodorant that lasts a 5-mile run" | h1 | "The deodorant that lasts a 5-mile run" | yes | MM1 |
| Offer | 15% off first order, code RUN15 | offer line + CTA href | "15% off your first order, code RUN15 applied" / ?discount=RUN15 | yes | MM2 |
| Product / variant | Magnesium stick, Cedar, 75 g, gid://.../456 | hero image + buy box default | same variant | yes | MM3 |
| Angle | no aluminium, 48 h | subhead | "48 hours without aluminium" (ledger P3) | yes | MM4 |
| Persona | runners | qualifier line | "Made for runners who sweat through the usual sticks" | yes | MM5 |
| Tone | UGC, first person | hero treatment | creator video poster, first-person subhead | yes | MM6 |
| CTA verb | "Get 15% off" | primary CTA | "Get 15% off my first order" | yes | MM7 |
| New claims above fold | none | first screen | none beyond ad claims + P1, O1 | yes | MM8 |
| Awareness | problem-aware (cold Meta) | page type | ad-landing-page | yes | MM9 |
| Visual | product on white, cedar tone | hero slot A1 | creative asset a_91f... | yes | MM10 |
```
## 5. Search-intent match (Google search and shopping)
| Id | Rule | Sev | Tag |
| --- | --- | --- | --- |
| MM14 | The h1 contains the query's head term (the product or category noun) and its modifier ("best", "vs", "for oily skin", "under ₹2,000") in the same sense. Use dynamic text replacement for the verb only when the ad group's verbs differ. | FAIL | RESEARCH |
| MM15 | The first screen at 390 answers the query: informational queries get the answer or the table first (`answer-first`, `comparison`, `seo-buyers-guide`); transactional queries get price, availability and CTA first (`pdp`, `pdp-hybrid-landing`). | FAIL | RESEARCH |
| MM16 | Query modifiers become sections: "vs" needs `comparison`; "best" needs `methodology` or ranking criteria; "review" needs `reviews` from the ledger; "price" or "cost" needs `pricing` above the fold; "near me" or "delivery" needs `shipping-returns` early. | FAIL | OPERATOR |
| MM17 | Shopping-ad clicks land on the exact product and variant in the feed at the feed price; a price change on the page without a feed update is a mismatch (and a Merchant Center disapproval risk). | BLOCK | LAW, OPERATOR |
| MM18 | The page title, meta description and h1 agree on the head term; the ad's display path words appear on the page. | WARN | RESEARCH |
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |