Rufus’s Pet-Supplement Query Rewrites Are Getting Weirdly Specific, The Catalog Work That Captures the New Traffic

11.4%. That’s the share of pet-supplement impressions across our 6 client accounts that came from query strings nobody typed in 2024. Rufus rewrote them. The convert rate on those rewrites is 2.7x category average. Most agencies are missing the entire mechanism.

The query “best joint supplement for senior labrador with hip dysplasia and sensitive stomach” did not exist in any keyword tool we use as of January 2026. By March it represented 0.4% of impressions on a single ASIN. Multiply that across hundreds of variants and the floor of a category shifts. We have been pulling Search Term Reports weekly since November and the pattern is now stable enough to act on.

This piece walks through what the rewrites actually look like, why traditional keyword harvesting cannot capture them, and the four catalog moves that determine whether your listing is the one Rufus picks. We will show numbers from real accounts and we will tell you which moves did not work.

What Rufus is actually doing to pet-supplement queries

Rufus does not match strings. Rufus interprets shopper intent and rewrites the query into a structured retrieval prompt. For pet supplements specifically, the rewrites cluster around five intent dimensions: species (dog, cat, horse, exotic), life stage (puppy, adult, senior, geriatric), breed sensitivity (small breed, large breed, brachycephalic), health condition (joint, anxiety, GI, skin, cognitive), and consumer constraint (grain-free, soft chew, picky eater, allergen-free).

A shopper types “joint supplement for my old dog.” Rufus rewrites that into a multi-dimensional retrieval prompt that scores listings on how completely they answer joint + senior + dog + chewable + safety. The listing that wins is not the one with “joint supplement” in the title 4 times. It is the one whose A+ content, bullets, and backend resolve all five dimensions clearly.

We pulled 142 unique long-tail query strings off three pet-supplement accounts in the first two weeks of April. 89 of them were not in Helium 10, Jungle Scout, or Brand Analytics. They converted at 6.3% blended versus 2.4% category average. They are real traffic. They are coming from Rufus.

Why keyword harvesting cannot solve this

Every agency we compete against still pitches “comprehensive keyword research” as a core deliverable. We did the same thing in 2022. It worked then. It does not work now, and the reason is structural.

Keyword tools scrape Amazon’s autocomplete, third-party search panels, and historical search-volume data. None of those sources see Rufus rewrites. Rufus rewrites are generated at retrieval time, per shopper, and they never appear in autocomplete because they were never typed. The Search Term Report is the only place they surface, and only after they fire enough impressions.

So the agency that is still selling keyword research is selling 2022’s solution to 2026’s problem. The serious work is downstream: ensuring your catalog actually answers the rewritten query in a structured form Rufus can parse. That is a different deliverable, with different leverage, and most agencies do not know how to scope it.

For deeper context on the pet-supplement category trajectory we modeled at the start of this year, read our pet supplements 2026 H1 data review. It is the baseline we are now comparing the Rufus shift against.

The four catalog moves that capture Rufus traffic

We tested twelve catalog modifications across six pet-supplement ASINs between February and April. Four moved the needle materially. The other eight did not. Here are the four that worked, in order of impact.

Move 1: Multi-dimensional bullet structure. Stop writing bullets that pile feature claims. Start writing bullets that resolve one intent dimension each. Bullet 1 = species and life stage. Bullet 2 = primary condition addressed. Bullet 3 = consumer constraint solved. Bullet 4 = format and palatability. Bullet 5 = safety and sourcing. We rewrote a top-selling joint chew this way in February. Sessions held flat. Conversion rose 23% in 6 weeks. The lift was almost entirely on long-tail rewrites.

Move 2: A+ comparison modules that name conditions. Rufus reads A+ content. Specifically it reads the comparison and feature modules. We replaced a generic “Why our chews are different” module with a condition-specific feature grid (hip dysplasia, arthritis, post-surgery recovery, daily preventive). Click-through to the buy box from rewritten queries rose 31% on that ASIN.

Move 3: Backend keyword fields rebuilt around intent verbs. Most backend fields are stuffed with synonyms (joint, hip, mobility, flexibility, etc.). That is keyword-matching thinking. We rebuilt three accounts’ backend fields around intent verbs and qualifier strings: “supports senior dog mobility,” “for picky eater small breed,” “grain-free joint chew.” The string-density approach lost. The intent-density approach won. Indexed coverage on Brand Analytics-rare queries doubled.

Move 4: Q&A seeding with shopper-voice questions. Amazon’s Q&A surface is now a Rufus retrieval source. Brands that seed authentic, shopper-voice questions (“Will this help my 12-year-old cocker spaniel with stiff back legs?”) and answer them factually pull retrieval weight that keyword-stuffed alt text cannot. We seeded 15 Q&As per ASIN across four products in March. By mid-April the affected ASINs were appearing in Rufus voice-style query rewrites at roughly 1.4x their pre-seed rate.

What did not work, and why we are telling you

Eight of our twelve tests produced no measurable lift. We are listing them because the absence is informative.

Title rewrites with more keyword density did nothing. Title length over 180 characters hurt mobile CTR. Adding more images to the gallery without changing alt text did nothing. Replacing a vector infographic with a photo of the product did nothing. Adding a video module did nothing for organic; it helped sponsored-display retargeting only. Rewriting the brand story module did nothing measurable. Adding more reviews-request friction (insert cards) did not change review velocity. Boosting PPC bids on long-tail rewrites did not improve organic ranking on those same rewrites.

The pattern: surface-level optimization is dead. Structural intent-resolution is the live work.

How to scope the work for the next 90 days

If you manage a pet-supplement brand and you want to capture Rufus rewrite traffic in Q3 2026, here is the sequencing we are running for our clients. Week 1: pull 90 days of Search Term Reports and bucket queries by the five intent dimensions. Week 2: identify the dimensions your catalog is weakest on. Week 3: rewrite bullets and backend on your top three SKUs to resolve each dimension cleanly. Week 4: replace one A+ module per SKU with a condition-named comparison grid. Week 5-6: seed Q&A in shopper voice. Week 7-12: monitor Search Term Report drift and iterate.

The accounts we have run this sequence on are showing 14-28% conversion lift on rewrite-driven sessions. Total session counts have stayed roughly flat or risen modestly. The lift is not from more traffic. It is from being the listing Rufus picks when it has a choice.

This is the work. It is not glamorous, it does not produce a dashboard screenshot you can post on LinkedIn, and it cannot be productized into a one-time audit deliverable. It is monthly catalog hygiene with a Rufus-specific lens. The agencies that figure this out in 2026 will eat the agencies that are still selling keyword harvesting in 2027.

If you want to see how your pet-supplement catalog scores on the five intent dimensions, book a 30-minute Rufus readiness review with ClearSight Consulting. We will pull your last 90 days of search term data and tell you which dimensions your listings are leaving on the table. No deck, no upsell, just the numbers.