From cold start
to first fix.
Four steps, in order. The first three you can do right now with no install, no signup and no card — start at step one and you will have a scored report inside a minute.
Point it at a store
Paste any Shopify storefront URL into the scanner. It reads the public endpoints a crawler reads and returns a scored report at a shareable URL.
Read the four numbers
One composite legibility score, and the four pillar scores underneath it that move it. Every point traces to a published criterion.
Work it in impact order
The report lists failures heaviest-first. Fix the top item, re-scan, and watch the composite move — measured after the change, not promised before it.
Install on the store
Installing the Shopify app turns the manual re-scan into a nightly pass over the full catalogue. Not live yet — see step four.
Step one — run the free scan
Open the free scan and enter a storefront domain — yours, or a competitor’s; the scan needs permission from neither, because it reads only what the store already serves to any visitor. There is nothing to install and nothing to approve.
What it fetches is the same short list an AI crawler fetches: robots.txt, sitemap.xml, llms.txt, products.json, your collections and policy pages, the raw HTML of a few product pages, and the response headers those come back with. It does not log in, does not touch your admin, and does not write anything anywhere.
When it finishes you land on a report at a URL you can share — send it to a developer or an agency and they see exactly what you see. A completed scan is re-used for 24 hours, so a second look does not re-run the crawl.
Step two — read the report
The number at the top is the composite legibility score. It is a blend, not a verdict: 11 scored criteria under rubric v1.0, weighted, and split across the four pillars in the fixed order this whole product uses — SEO, AEO, GEO, AIO. A pillar score tells you which discipline is costing you; the criteria beneath it tell you which specific check failed and what it was worth.
Two things are scored separately from catalogue quality, and it is worth knowing that before reading a low number as a catalogue problem. Crawler access is one: a blocked AI agent is a hard exclusion rather than a ranking penalty, so if your robots.txt shuts out any of the 10 named agents, that is a gate and not a grade. Structured data is the other — it is read from the rendered HTML of product pages, so a theme that strips JSON-LD can drop that score without a single product changing.
Nothing in the report is an opinion you have to take on trust. Every criterion and weight is published, so a score you disagree with is a rule you can go and read.
Step three — fix in impact order
Failures are listed heaviest-first, because the order is the advice. Working down a list of twenty small gaps while the heaviest one stays open is the most common way to spend a week and move the composite by nothing.
Each failure carries the class of fix it needs, and the classes are not interchangeable. Some are automatic — resolvable from data already verified in the record, and the kind the app writes for you once it is connected. Some are assisted: the engine can draft it, but it asserts a fact only you can confirm. Some are yours outright — a GTIN, a material, a shipping fact that exists nowhere but in your head. And some are guided: work outside the app’s three scopes, such as a theme change, where you act and we tell you exactly what to change.
Fix the top item, then re-run the scan on the same domain. The composite you see afterwards is measured against the changed store, which is the only version of a before-and-after worth anything.
Step four — connect the store
This step is not live yet. The Shopify app is built against the flow below and is not on the App Store, so no merchant has installed it and no store has ever been written to. It is set out here because it is the next step and you should know what it will ask for — not because you can take it today. The changelog records what has actually landed, and its last section is the honest half.
The install is an OAuth handshake on Shopify’s own consent screen — no password to share, no staff account to create, no developer access to grant. Three scopes are requested: read_products, write_products and read_inventory. Deliberately absent is permission to edit your theme, which is why enabling the storefront JSON-LD block is something you switch on yourself rather than something we do silently inside your layout.
Once connected, the manual re-scan of step three becomes a nightly pass over the full catalogue, and automatic fixes are written into one dedicated metafield namespace. Every other mutation — including the ordinary product update most apps in your admin use — is refused at the API client before the request is issued. Your titles, descriptions and copy are never a write target. How that contract is enforced.
You can revoke the scopes from your Shopify admin at any moment, and uninstalling does the same thing. Because writes are confined to that one namespace, removing them cannot touch a field you wrote.
If you get stuck
The FAQ answers what comes up most — what the scan checks, what the app will and will not write, how pricing is banded, and where the scoring is published. The documentation index lists every reference page in one place. If neither has it, write to hello@ranksniperhq.com or use the contact form — a human answers.
