A launch checklist, not a schema
The fields that make the range choosable, separated from the fields that complete the record. Both matter eventually; only one of them blocks a date.
Use case
A new range is not blocked by how much data you have. It is blocked by not knowing which fields a buyer actually chooses on, so the plan becomes enrich everything or ship blind. The audit gives you the minimum set that makes a range decidable, measured before launch, with no traffic needed.
Fields in scope before launch
The rest are not abandoned, they are sequenced. The difference between those two numbers is whether the date moves.
Where you are
Fields the schema allows for this range
200
Supplier data comes in as spreadsheets, PDFs and feeds, each with its own idea of what a field is called and which ones are mandatory. Somebody has to turn that into a catalogue, and the honest estimate for doing it properly is longer than the window.
So the range launches in one of two bad states. Either it waits for a complete enrichment pass and misses the season it was bought for, or it goes live with names, prices and images, and no attribute a shopper can filter on. The second is more common because the date is real and the data quality is negotiable.
A range that launches unfilterable is worst exactly when it matters most. The launch window is when interest is highest and when you have the least behavioural data to fall back on, so the catalogue has to carry the whole decision on its own.
Enriching faster still means enriching everything. The question nobody answers before a launch is which twenty of two hundred fields are actually load-bearing.
The cut line nobody draws
A launch does not need a complete schema. It needs to know which fields are load-bearing on day one.
All three are the same team and the same quarter. Only the last one is chosen rather than guessed.
What it looks like
The full enrichment plan against the subset that makes the range shoppable on day one.
Both end in the same place. Only one of them ends before the date.
What happens
Not by doing the same work faster. By doing a smaller piece of it first.
Send an export of the incoming products, however incomplete. The measurement works on partial data, which is the point: it is telling you what is missing, so it cannot require you to have already filled it in.
The handful of attributes that carry most of the decision in that category, ranked, with how much each one moves the range from indistinguishable to choosable. That is your launch checklist, and it is usually far shorter than the schema.
Re-measure as the rest lands, so the backfill is prioritised by what is still costing you rather than by what is still empty. The range gets better in a visible order instead of asymptotically.
What you get
The fields that make the range choosable, separated from the fields that complete the record. Both matter eventually; only one of them blocks a date.
Extractable from supplier titles and descriptions, mappable from a feed you already receive, or genuinely absent. That split decides what is possible before the date and what is not.
Decision readiness is computed from the catalogue, so a range can be assessed before a single visitor sees it. Waiting for behavioural data means finding out after the window closed.
"We are launching with these nine fields and backfilling the rest" is a position you can take to a buyer or a merchandising director, because the nine were chosen by measurement rather than by whichever columns the supplier happened to fill.
Before you commit
Questions
One export of an incoming range, and a ranked list of the fields that decide whether it launches shoppable or launches invisible.
Request a Decision AuditIt runs on incomplete data. That is the state it is built to assess.