When we wrote that catalog health is upstream of ads, the response we kept getting was: what does that mean, in modules? So here’s the rebuild, the actual A+ work we do on every new account in the first 30 days. Six steps. Two weeks until you can read the result in the ads dashboard. It’s the most concrete piece of the bigger thesis: your catalog is now a software product.
This isn’t a creative refresh. The visuals on most of the accounts we audit are fine. The problem is structural, the modules are doing the wrong jobs in the wrong order for an algorithm that’s now reading them differently. Most A+ stacks were designed in 2020-2022, when the page was scored on keyword density and image quality. Cosmo doesn’t score that way. Cosmo reads the page as a document and decides whether the document answers the rewritten query. Two pages with identical bullet points and identical images can perform completely differently if one of them sequences the modules in a way Cosmo can parse and the other doesn’t.
1. Re-anchor the H1 module to the rewritten query, not the brand
Most A+ stacks open with a brand-banner module. In 2020, that was correct, recognition was a ranking signal. In 2026, with Cosmo doing semantic match against the page, the first module is real estate you cannot afford to spend on logo recognition. Replace it with a problem-statement banner that uses the language of the rewritten query, not the keyword you’re bidding on, but the question the customer is actually asking.
Pull these from your ST report’s “customer search term” column where conversion-share is highest. The actual phrasing matters. “Best knife sharpener for arthritic hands” converts at 3.1x our category average; the brand whose A+ opens with “Designed for hands that ache after a long day” gets the assist. The brand whose A+ opens with “Premium kitchen tools since 1987” gets nothing. We’ve watched the gap between those two openers move 4-7 points of CTR on identical bid structures.
2. Strip the lifestyle module that isn’t doing semantic work
You probably have a “lifestyle” module showing the product in use. If it’s there for vibes, it’s costing you. Replace the body copy under it with explicit attribute-completeness language, material, dimensions, use case, written in full sentences (not specs). Cosmo reads sentences. It doesn’t read your bullet structure as well as the agency that built your last A+ told you it does.
The test we run on every lifestyle module: cover the image, read just the body copy aloud. If a buyer can’t tell what the product does or why it would solve their problem from those words alone, the module isn’t doing semantic work. It’s a placeholder. The fix is usually 80-120 words of plain prose under the existing image, listing what the product is, what it’s made of, and what kind of buyer should pick it. We’ve seen detail pages move from a 9% organic-share floor to 14% on this single module change, with no bid changes.
3. Comparison chart, but as a substitution matrix
The standard “compare to other products in our line” chart is a conversion module, not a discovery module. It helps the buyer who’s already on your detail page pick between your SKUs. It does nothing for the buyer who landed there by accident from a query you’d never bid on. Replace it with a substitution matrix: “If you came here looking for X, here’s why this is the right pick over Y.”
This is what Rufus is doing in its summaries; you may as well source the language from the same place. We seed the substitution matrix from the actual queries that brought traffic to the page (from your ST report’s bottom 30%), and from the competitor SKUs that buyers most often comparison-shop against (from Brand Analytics’ “Comparable Products” panel). Three to five rows is plenty. The point isn’t a complete competitor map, it’s a Cosmo-readable answer to the question the buyer was actually asking.
4. Insert a FAQ module and write it for voice and AI search
Three FAQ items, each 2-3 sentences. Pull the questions from the actual long-tail queries in your ST report, the ones that converted at 2x category average. This is the module that earns you Rufus citations. It’s also the one most agencies skip.
The structure that works: question phrased as the buyer would actually phrase it (not “What is the warranty?” but “Does this come with a warranty if I’m using it commercially?”), then a 2-3 sentence answer that’s specific enough to be quoted. Generic answers get nothing. Specific answers, actual numbers, actual edge cases, actual conditions, get pulled into Rufus summaries and become impressions you didn’t pay for. The FAQ module is the lowest-effort, highest-leverage A+ change we make on most accounts.
5. Move social proof to the bottom, not the top
Reviews and awards belong below the substitution matrix, not above it. Cosmo doesn’t index them as relevance signals. Buyers who needed them have already scrolled, they made the relevance decision earlier on the page, based on the first three modules. Putting badges and “as seen in” logos at the top of the stack burns the most valuable real estate on a signal that doesn’t change Cosmo’s ranking and doesn’t change a buyer who already has intent.
Move them, but don’t kill them. Social proof still matters at the conversion-decision moment, for the buyer scrolled to the bottom of the page who’s ready to add to cart but wants one last reassurance. That’s where badges earn their place. Not at the top.
6. The mobile re-cut
Top 60% of the A+ stack on mobile carries the post. Anything that doesn’t survive a phone-screen recut needs to be either rebuilt or deleted. We delete more than we rebuild. The Premium A+ modules that don’t render well on mobile are usually subtracting, not adding.
The audit we run: take the same A+ stack you just rebuilt and view it on a 375px-wide phone screen. If a module’s headline is illegible without zooming, it’s failing. If a comparison chart drops to a vertical scroll that takes 4 swipes, it’s failing. If your “lifestyle” image becomes ambiguous when shrunk, it’s failing. The mobile re-cut isn’t a separate workstream, it’s the test that decides whether each desktop module ships.
What changes downstream
Two weeks after this rebuild lands, most accounts see CTR on Sponsored Products climb 6-12% on the same bids. That’s the catalog change reading through to the ads, same impressions, better-matched intent, higher click-through. It’s not magic. The detail page is finally answering the query that brought the impression in.
The CTR lift is the early signal. The bigger move comes in week 4-6: organic-share starts climbing as Cosmo’s ranking model picks up on the improved relevance signals. We typically see 8-15 points of impression-share recovery on the queries we targeted with the rebuilt modules. That’s recovery, not growth, most of the share you reclaim was yours to begin with, before Cosmo’s reranking moved it elsewhere.
And once the page is answering correctly, your negative-keyword harvesting starts catching real waste instead of fighting catalog drift. Half the keywords that looked wasteful before the rebuild were catalog-symptom, the page couldn’t convert them. After the rebuild, those keywords either convert at the new floor or get killed cleanly. The signal-to-noise ratio in the ST report goes up, which compounds back into the ads program.
Get an audit, we’ll show you which modules are working against you.