Decision Audit
Send one category export. In 48 hours you get a ranked list of the attributes worth recording, the products missing each one, and where each missing fact already lives. No code, no tag, no customer data, and nothing to install.
What this shopper already knew
2/3of what they knew is not a field, so this visitor is shown the same order as everyone else.
Where you are
Fields your schema could hold
200
That was never the hard part. A catalogue has hundreds of possible fields and nobody can tell you which ones to start with, so the work either never begins or it begins everywhere at once.
Your completeness dashboard stays green throughout, because it measures whether the fields you defined are filled, not whether they were the right fields.
You need to be told which twenty fields out of two hundred to do first, and which of those you already own.
Why nobody else can tell you this
A shopper leaves without buying. Working out why is the whole business, and there are only two ways to do it.
Arrived
from a price comparison page
Searched
“lightweight”
Filtered
nothing to filter on
Chain breaks hereOpened
three products, then two more
Left
without choosing one
The session did not fail at the checkout. It failed at step three, because “lightweight” is not a field in this catalogue, so nothing could narrow and the shopper was left comparing by hand. Filtering is where most shelves break: 78% of mobile ecommerce sites score poor or mediocre on it, and 58% on desktop. Baymard Institute, 2025 benchmark
Click and conversion models weigh the fields that already exist. A field nobody recorded generates no clicks, so it scores zero and stays invisible no matter how much traffic you send at it.
Needs months of traffic. Never surfaces the missing field.
Where the chain breaks is a property of your catalogue, not of your visitors, so it can be measured from an export. That is why this works on a new category, a new market or a launch range with no history at all.
Needs one CSV. Works on day one, with zero traffic.
What it looks like
Ranked by bits of decision information, with the affected products attached to each row.
Ranked by bits of decision information, not by how much is absent.
Not our numbers
What happens
Three things happen, and only the first one needs anything from you.
A CSV or JSON export of a single category. Product name, description, category, price, and whatever attributes you already record. Pick the shelf where you most suspect shoppers struggle.
We work out how much each attribute contributes to telling your products apart today, and how much it would contribute if it were recorded everywhere. The difference is what that field is worth, measured in bits rather than guessed from clicks.
You get the ranked list and the working files, then half an hour with whoever owns merchandising to go through what it found and what it would take to act on it.
What you get
Six things, all usable by someone who was not on the call.
Every attribute, ordered by how much recording it would improve your shoppers' ability to choose.
Not a percentage. The actual SKUs, as a working file your team can filter and assign.
Extractable from your own text, mappable from a supplier feed, or genuinely absent.
How many products a shopper still cannot separate. Re-measurable after the repair.
On the shelf we measured, 161 of 216 values were in the merchant's own titles.
With whoever owns merchandising. If your catalogue is fine, we say so on that call.
Before you commit
Questions
One category, a fixed fee, and an answer in two days that is useful whether or not you ever work with us again.
Request a Decision AuditNo obligation at any step, and no integration at any point.