Your customers are asking
questions your catalog
can’t answer.
Paste your store URL. In about a minute we’ll show you which questions fail — using your real products.
Free. No install, no login, no Shopify permissions. See a sample report
of these 6 questions returned nothing. This store stocks products that should have matched all six.
A better search engine can’t find an attribute you never wrote down.
Semantic search, vector search, AI search — they all do the same thing. They rank the data in your product records.
So when a shopper asks for the one that fits a 2019 F-150 with the 3.3L and your catalog doesn’t store engine variant, there is nothing to rank. The engine didn’t fail. It reported.
Most catalogs describe what a product is: name, price, size, colour, a paragraph of marketing copy. Almost none describe what it is for, what it fits, or what it solves.
What the shopper asked
“waterproof hiking boots, wide fit, under £150”
What your catalog stores
title · price · colour · size · vendor
No width fitting. No waterproof rating. No activity.
What any engine can return
Nothing to match on.
Most catalogs fail in the same places.
We’ve checked N Shopify catalogs against the questions real shoppers ask in each store’s category. The median store answered X out of 40.
Placeholder distribution. Do not publish until N ≥ 50 real audits.
Third-party evidence paragraph — cite Baymard search UX benchmark and the Klevu natural-language retailer study.
See where your catalog landsOne gap. Four places it costs you.
Every system that has to find a product for a shopper reads the same records. When those records are thin, all four degrade at once — and only one of them tells you.
On-site search
Need-based queries return nothing, or return the wrong thing. This is the only failure you can currently see, which is why most merchants think it’s the whole problem.
0 results
Collection filters
Shopify builds storefront filters from your product options and metafields. Empty fields mean a large collection with two filters on it and a shopper doing the sorting themselves.
2 filters · 4,000 products
Shopping feeds
Missing attributes get listings suppressed or under-served in Google Shopping and marketplaces. Your ads compete on data quality before they compete on bid.
812 items · limited performance
AI shopping channels
Shopify now syndicates eligible merchants’ catalogs to AI assistants by default. Ranking factors can include data quality and relevance.
syndicated: what you had
Dated urgency line — state the Shopify Agentic Storefronts default-on date as fact, cite Shopify.
source link
You can’t see this loss in your analytics. That’s what makes it expensive.
A zero-result search gets logged. Almost nothing else does.
The shopper who opened your filters, found three of them, and left. The listing that didn’t serve. The AI assistant that recommended someone else. None of it appears in Shopify Analytics or GA4.
Your conversion rate is calculated on the traffic your catalog managed to survive. It looks fine because the shoppers it failed were never counted.
Vertical-specific cost line. For parts merchants, use the published fitment-return figures and cite the source.
source
Find the gap. Fill it. Prove it moved.
We read your catalog the way a shopper does
Not a checklist of empty fields — a retrieval test. We run your actual products against the questions your category actually gets asked, and find where the catalog can’t answer.
We infer the missing layer
Attributes and relationships — use case, compatibility, suitability, substitution — derived across your whole catalog, not guessed one product at a time. Anything below our confidence threshold is flagged for you, not written.
We write it back into Shopify
Structured metafields and taxonomy attributes, in your store. Search reads it. Filters read it. Your feeds read it. Then we measure whether it moved anything.
product 4412 — today
product 4412 — after
Note: the arrow values are reference-type metafields pointing at other products.
Everything we infer gets written into your catalog.
Not into an index you rent access to. Into your Shopify product records.
If you cancel ReLUnit tomorrow, the attributes stay. Your filters keep working. Your feeds keep serving.
That’s deliberate, and it’s the main reason a search vendor won’t do it — their index is what keeps you. Ours doesn’t have to.
SCREENSHOT: real Shopify admin, product page, ReLUnit-written metafields populated. Not a mockup.
Where ReLUnit fits alongside what you already run.
A search app
Ranks the data you already have. If the attribute isn’t in the record, there’s nothing to rank. Keep your search app — this is what makes it work.
A PIM
Gives you a governed place to store attributes. It doesn’t produce them. Under 50,000 SKUs on Shopify, you probably don’t need one yet.
Shopify’s native tools
Define roughly a thousand category attributes for you and add the fields. They don’t fill them in, and they infer one product at a time.
A bulk AI editor
Fills one field at a time on one product at a time. That’s the easy half. Knowing that this part fits that vehicle is the half your customers ask about.
An agency cleanup
A snapshot. It decays with the next supplier feed.
ReLUnit isn’t for you if you have a few hundred SKUs, a hand-curated single-brand catalog, or products people find by name.
Shopify’s native filtering will do the job and you should keep your money.
A 90-day pilot, against a number we agree on first.
Before we start, we agree the target: catalog answerability on your category’s query set, attribute coverage on the fields your surfaces depend on, and zero-result rate.
If we don’t hit it in 90 days, you don’t pay for the pilot.
We also measure revenue effect against a holdout and report it honestly — including when it’s inconclusive. Attribution over ninety days on a live store is noisy, and we’d rather tell you that than sell you a number we can’t defend.
Pilots start at $X/month and are scoped after your audit.
Objections, answered plainly.
Shopify’s taxonomy defines the fields — over a thousand category attributes — and adds them to your products. It doesn’t fill them for your catalog, and it infers one product at a time from what you already gave it. Shopify Catalog also cleans and syndicates whatever data you have. That’s the point: whatever you have.
Keep it. We’re not a search engine. We feed the one you’ve got, plus your filters and your feeds.
Confidence thresholds, review queue, reversibility. Be specific about the threshold and the review step — this is the top objection in regulated categories.
Not without you approving it. The audit is read-only and doesn’t require an install.
It stays in your Shopify catalog. That’s the design.
The audit takes about a minute. Write-back takes days. Measurement takes the ninety.
Concrete answer: what you store, for how long, who can see it, what scopes the app requests. Not "security is our top priority".
You can find out in about a minute.
Paste your store URL. We’ll show you what your catalog can’t answer.
Check your catalogFree. No install, no login, no Shopify permissions.
Talk to us about a pilot
Tell us a bit about your store and we’ll follow up to scope a 90-day pilot.