Your store is invisible to the robots now doing the shopping.
AI shopping agents skim your store in milliseconds, miss half of it, and hand the sale to a competitor. ShelfGrader reads your store the way ChatGPT, Gemini and Perplexity do, then gives you a letter grade and the exact fixes.
yes, this means yours too
an agent picked Wander Goods instead
Go on. Roast my store. · 10 signals checked · we never touch your live site
A shopper sees a store. An agent sees this.
An agent does not look at your photography or your fonts. It parses what is machine-readable, fills a shortlist, and moves on. If the data is not there, neither are you.
<title> Home | Bright Little Studio
name ............ "Candle"
price ........... (not found)
availability .... (unknown)
schema.org/Product none
rating .......... (unreadable)
alt text ........ ""
llms.txt ........ 404verdict: not enough to recommend.
moving to next store.
it can't read what it can't parse. so it leaves.
The 10 machine-readable signals we grade.
100 points totalProduct schema /16
Valid Product JSON-LD with name, price, availability and brand.
Machine-readable price /12
Price and currency in structured data, not baked into an image.
Title and meta /8
One clear title and meta per product. No "Home | Untitled".
Description depth /8
Specifics an agent can match: materials, size, use, who it is for.
Images and alt text /8
Real alt text on product images. Agents read words, not pixels.
Reviews and ratings /10
Rating value and review count marked up so agents can cite them.
Availability and shipping /10
In stock, lead time and shipping in schema, not just a banner.
Trust signals /8
Returns, guarantees and contact details an agent can find.
Agent directives /10
An llms.txt that tells agents what to read and what to trust.
Crawlability /10
Agent crawlers allowed; product pages reachable without JS gymnastics.
Questions, answered.
Q.Is this actually free?+
Yes. The grade, your ranked fixes, and the competitor an agent picks instead all cost nothing. No card, no signup, no catch. We make money only when you decide to fix things.
Q.Do I need to be technical?+
No. If you can paste a URL, you can get graded. The fixes are written for humans, and the paid packs do the technical part for you.
Q.Will this change or slow down my store?+
We never touch your live site to grade it. We read the same public pages an agent reads. The fix pack gives you code to paste in when you are ready, on your schedule.
Q.Which agents do you check for?+
We test the way ChatGPT, Gemini and Perplexity shop: fetch the page, parse what is machine-readable, and build a shortlist. If they cannot read you, you are not on it.
Q.I got an F. Is my store doomed?+
Good news: an F is mostly fixable in an afternoon. Missing schema and a missing llms.txt sink most scores, and both are paste-in fixes. Re-grade when you are done.
Q.What do I get for $20 versus $99 a month?+
Twenty dollars is a one-time pack: we generate your schema, price markup and llms.txt, and you paste them in. Ninety nine a month is done for you: we set it up and keep it current as your catalog changes, with monitoring and re-grades. Cancel anytime.
Three ways to pass.
The honest report card so you know exactly where you stand.
- +Letter grade out of 100
- +Your ranked fixes
- +The competitor agents pick
- +No signup, no card
We write the code. You paste it in. Watch the grade jump.
- +Your product schema, generated
- +Price and availability markup
- +A ready-to-paste llms.txt
- +A free re-grade to prove it moved
Done for you, and kept that way as your catalog changes.
- +Everything in the fix pack
- +Set up and kept current for you
- +Monitoring and monthly re-grades
- +Alerts when your score slips
- +Cancel anytime