Brief sections
The leak
A manufacturer sends the new price list — a PDF, naturally, formatted differently than last year’s. Someone spends three days keying it into the system, the website still shows old prices for two more weeks, and a customer orders at a price that no longer exists. Multiply across every line you carry: descriptions written by four people over ten years or copied straight from vendor boilerplate, options that exist in the price book but not on the website, and a version-control strategy best described as a prayer. The cost shape is quoting errors, channel inconsistency, and an army of small manual updates that never ends.
What AI realistically does here
Bulk work with judgment sampling. It ingests vendor price lists and spec sheets in whatever format they arrive, maps them to your catalog, and stages the updates with every change diffed for review. It generates and maintains descriptions to your voice rules, checks consistency across channels — site, price book, marketplace listings — and flags anomalies: the price that dropped by half overnight (probably a keying error upstream), the SKU that exists in three places under two names. Cab‑O‑Matic, our pricing and quoting product for cabinet dealers and homebuilders, has made the lesson clear: at catalog scale, the hard part is mapping and reconciliation, not generation.
What has to be true first
- One master source for the catalog, even if it’s currently a spreadsheet. If the site, the price book, and the ERP disagree, we pick the winner per field before automating anything — otherwise you get faster inconsistency.
- Someone with authority to adjudicate conflicts, because there will be conflicts, and they’re business decisions dressed as data problems.
- Voice and format rules for anything customer-facing: two paragraphs or two lines, spec-first or benefit-first. The rules can be simple; they have to exist.
What we'd install
Catalog work gets scored in the opportunity portfolio like everything else — it usually ranks high when it feeds quoting, because then one fix drains two leaks. The reconciliation rules and voice guidelines go into the SOP library, the catalog agent runs on a schedule with a named owner reviewing diffs, and the value ledger tracks update lag — price-list arrival to live-everywhere — and errors caught before they shipped. Sequence on the AI operating partnership.
Worth it?
Add up the hours your last three price updates took end to end, and the cost of the last order that shipped at a stale price. The calculator below turns the recurring hours into an illustrative annual figure — illustrative math, not a promise. If your catalog has no master source at all, the scorecard will route you to fix that first, and it’s right.