How to read your SchemaCart report
What the score band means, what Good, Needs work, Broken and FYI mean on each check, why some checks are left out, and what to fix on your store first.
Your report has loaded. There is a number, a word next to it, a row of coloured labels, and a list of findings underneath. This is what each of those is telling you, in the order you meet them, and where to start once you have read them.
The word matters more than the number
Your score is out of 100 and lands in one of three bands.
- Readable, 70 or above. Your store gives AI tools and search engines most of what they need to make sense of it.
- Partly readable, 45 to 69. Some of what they need is there, and some is missing or unclear.
- Hard to read, below 45. Large parts of your store are difficult or impossible for software to make sense of right now.
Read this as a measurement, not a grade. It says how readable your store was at the moment we checked it, which is why the report carries the date it ran. It does not say whether any AI tool will mention your products, because nobody outside those companies decides that. What you do control is whether your store gives them something clear to read, and that is the thing this number counts.
Beside the score you will also find your store's address and the platform we detected, so you can confirm at a glance that you are looking at the right store. Every report gets its own private link, so it stays yours to share or to reopen later without running the scan again.
The four labels on each check
Every check ends with one of four labels.
- Good. Working as it should. Nothing to do.
- Needs work. Partly there. It earns half the points and is usually the cheapest thing on the page to finish.
- Broken. A real problem we confirmed. Zero points, and this is where effort pays best.
- FYI. Something worth knowing about your store that we do not score either way.
There is a counter near the findings showing how many of each you have. Read that row before reading any individual finding, because it tells you whether you are looking at an afternoon of tidying or a genuine project.
The two groups behind the score
Your score comes from eleven checks in two groups, kept separate on purpose so you can see where your points actually come from.
AI readiness is worth 75 points across five checks. Whether automated readers can reach your store at all is the heaviest single check on the page at 26 points. Your product details, written in a form software can read cleanly, are worth 19. The llms.txt file, a plain text file at your store's root that tells AI tools what your site holds and where, is worth 13. Content that answers the questions shoppers actually ask is worth 9, and page speed is worth 8.
Search basics are worth 25 points across six checks: your titles and descriptions, your sitemap, your secure address, having one main address per product, your heading order, and your image descriptions. Old ground for search engines, and the same groundwork AI tools read your store through.
A store can do well in one group and badly in the other, which is exactly why the report does not blend them into a single number. If your 75 point group is healthy and your 25 point group is not, you have an afternoon of small fixes. The reverse is a bigger conversation.
Inside one finding
Open any finding and you get the same four things in the same order.
First, what we actually observed, stated plainly, for example that a file was missing or that a rule turned our reader away. Then What this means, which translates that observation into a consequence for your store. Then Why it matters, which is the argument for spending time on it. Then Do this now, which is the specific change to make and the way to confirm it worked afterwards.
Some findings also carry an example file or snippet of code you can hand to whoever does the work. Where that appears, it is a generic example of the right shape, not something generated from your catalogue. The paid audit is where those get built from your store's real data.
Checks we could not read are left out, not counted against you
This is the part most worth understanding, because it is easy to assume an unreadable check is just a polite way of saying broken. It is not.
When a check cannot be completed, usually because something stood between our reader and your pages, we remove it from your score rather than counting it as a failure, and we work the score out again from the checks we could genuinely finish.
Say ten of the eleven checks completed and one could not be read. That one is dropped, and the remaining ten each carry slightly more weight than usual so the total still adds up to 100. Your score is built entirely from what we confirmed, and never from an assumption about the part we could not see.
We do this because a confirmed problem and an unseen one deserve different labels. If a security setting blocked part of the scan, that might mean nothing at all is wrong with your store, only that we were not allowed to look. Marking that as broken would be a lie that flattered nobody.
It does still deserve your attention, just a different kind. An excluded check costs you no points, and it is telling you that something stops automated readers partway through your store. Finding out what, and scanning again, turns it into a real answer either way.
Check your own store
If you are reading this before running a scan, the fastest way to see all of it laid out for your own store is to run one. It takes about 45 seconds, reads your store the way an AI tool would, and puts the full report on screen. No plugin, no card, and an email address is optional. Run the free scan.
What to fix first
With a page of findings, an order helps more than enthusiasm.
Broken results in the AI readiness group.
They carry the most points and they are confirmed problems, so they move your score furthest. The access check alone is worth 26 points.
Needs work results with the highest point value.
These are already half working, which usually means a smaller job than a Broken result for a similar number of points.
Excluded checks, by finding out why.
They cost you nothing directly, but they mean something is turning automated readers away, and that is worth knowing before you spend money on anything else.
Search basics, once the bigger picture is clear.
Individually small, collectively 25 points, and most of them are settings rather than projects.
Treat that as a starting order, not a rule. If a Needs work result is five minutes of your own time while a Broken one needs a developer and a fortnight, take the quick win first and feel fine about it. The only outcome the ordering exists to prevent is you spending a whole afternoon on a three point fix while a twenty six point failure sits untouched further down the page.
One report is one moment
A report is a snapshot. Themes update, plugins get added, security rules tighten, and your score can move without you having touched anything on purpose. If you would rather see that over time than remember to check by hand, that is what our paid monitoring is for.
For now, the free scan is the place to start. Run it, read the band, read the two groups underneath, then work down from the Broken results carrying the most points. Run the free scan and see your own report laid out exactly this way.