Stock & Availability Data Services

Because most stock-outs in fast-moving categories are short, so daily sampling misses them entirely and systematically understates the problem. It also cannot compute duration — two products flagged out of stock today may have been out for one hour and eleven days.

We capture transitions rather than states: stock-out start timestamps, running duration, event counts and cumulative out-of-stock minutes over trailing windows. That is what lets you size the commercial cost instead of counting flags.

Two separate fields, and it is the most consequential design decision here. listed says whether the product is ranged at that retailer or location at all; in_stock says whether it is buyable where it is listed.

Availability percentages are computed only across the listed population. Merging the fields turns a range gap into an apparent supply failure — we have seen brands escalate a 40% unavailability figure to supply chain and spend weeks discovering most of it was never-ranged locations.

Yes, as event records with timestamps and the duration of the gap they closed. Replenishment is as informative as the stock-out: recovery time distribution, replenishment cadence and repeat-offender patterns all come from it.

It requires continuous polling rather than sampling, which is why most availability datasets do not have it. On the hourly tier it is standard.

Where a retailer publishes it, yes — typically via a store stock checker or click-and-collect availability on the product page. Coverage varies enormously by retailer and we confirm the pattern per retailer before build.

Where availability is revealed only inside a basket flow, we do not collect it, because that means creating sessions and adding items. Every record carries location_scope so you always know whether a figure is store, area or national level.

No, and nobody scraping public pages does. Inventory depth, units on hand and allocation are internal retailer data.

Where a retailer displays a low-stock hint — "only 3 left" — we capture it exactly as displayed, as text or a number, without treating it as a verified quantity. Those hints are often thresholds rather than counts, and presenting them as inventory levels would be misleading.

We distinguish them where the evidence supports it, and flag the ambiguity where it does not. A product unavailable long enough that a delist is plausible carries a flag rather than a determination.

Continued observation resolves most cases: a delisted product typically disappears from the category listing entirely, while a stock-out usually remains listed as unavailable. For range analysis specifically, catalog and assortment data treats delisting as the primary signal.

Hourly where availability is the primary use case, several times daily otherwise. Daily is close to useless for transition capture — you will see that a product went out without seeing when, for how long, or whether it came back.

We scope which SKUs and locations justify hourly rather than applying it universally, since hourly collection across a full estate multiplies cost substantially for SKUs that rarely go out.

It can often explain them, and it is one of the more satisfying uses. Joining out-of-stock minutes per SKU and location to your own sales data usually accounts for a meaningful share of unexplained variance.

It also improves forecasting, because demand history suppressed by stock-outs is otherwise fitted as genuine low demand. Correcting for out-of-stock minutes is a straightforward adjustment once the duration data exists.

We quote individually. The dominant driver is locations multiplied by SKUs multiplied by frequency — store-level hourly collection is the most expensive configuration in our catalogue, and often unnecessary at full scope.

A defined SKU set at national online level with several-times-daily refresh sits at the lighter end. Store-level hourly across a large estate sits considerably higher. We design the location sample with you before quoting. One scoping call, a free pilot on your own SKUs within 48 hours, then a fixed monthly quote. Request a quote.