What Is Assortment Intelligence? – Definition and How It Works
Learn what assortment intelligence measures, why product matching is the hard part, and which metrics it produces.
Assortment intelligence is the analysis of which products a retailer or marketplace actually offers, compared with competitors. It answers what is carried, what is missing, and what is new — not what things cost.
1. What It Measures
| Metric | Question answered |
|---|---|
| Assortment size | How many distinct products are carried |
| Overlap | What share of a competitor's range you also carry |
| Exclusives | What only you, or only they, carry |
| Gaps | What competitors carry that you do not |
| Category depth | How many options within a category |
| Brand coverage | Which brands are represented, and how deeply |
| New arrivals | What has entered the range |
| Discontinuations | What has left it |
| Availability | What is carried but out of stock |
The gaps and exclusives metrics are what buyers actually act on, which makes them the point of the exercise.
2. Product Matching Is the Hard Part
Assortment comparison requires knowing that two listings are the same product. That is entity resolution, and it is where most of the effort goes.
The same product across two retailers:
A: "Acme Wireless Keyboard K450 - Black, Mechanical, US Layout"
B: "ACME K450 Mechanical Keyboard (Black) US"
Different titles, possibly different images, possibly different specification formats. Matching them needs identifiers where available and inference where not.
Reliable identifiers — GTIN, EAN, UPC, ISBN, MPN. Where present, matching is nearly solved.
Where absent — brand plus model number extraction, specification comparison, image similarity, and title similarity after normalisation. Each is imperfect; combined with a threshold and a review band, they are workable.
3. Variants Multiply the Problem
Is a keyboard in black and one in white the same product or two? Is a three-pack the same as a single unit?
There is no universal answer, and the decision must be explicit and consistent, because it changes every metric. A retailer listing each colour separately appears to carry three times the assortment of one using a single variant-selecting page.
Normalising variant handling before comparing is what makes assortment counts comparable at all.
4. Collection Requirements
Assortment work needs coverage, which is a different requirement from price monitoring's freshness.
- Complete category traversal, not a sample. A missing page is a false gap.
- Pagination handled to the end, with verification that the last page was reached.
- Facet and filter combinations where a category is only fully enumerable through them.
- Location awareness, since assortment varies by region and store.
- Availability captured separately from listing existence — carried-but-out-of-stock is not the same as not carried.
The false-gap risk is the one to design against: an incomplete crawl reports a competitor as not carrying something they do carry, and a buyer acts on it.
5. Distinguishing Absence From Failure
A product not found can mean it is not carried, or that collection failed. These produce opposite business decisions, so they must be distinguishable.
Practical measures: record per-category item counts and compare against the site's own stated totals, alert on a category returning substantially fewer items than the previous run, treat a zero-result category as a failure until proven otherwise, and require two consecutive observations before reporting a discontinuation.
6. What the Analysis Supports
- Range planning — which gaps represent genuine demand.
- Competitive positioning — where your range is deeper or thinner.
- Brand negotiation, with evidence of comparative coverage.
- New-product detection, so competitor launches are visible early.
- Category trend detection, as assortment shifts precede demand shifts.
- Supplier assessment, showing which suppliers distribute widely.
New-product detection is often the highest-value output, because it is time-sensitive and there is no other route to it.
7. Assortment Versus Price Intelligence
They use the same collected data and answer different questions.
Price intelligence needs high-frequency observation of a known product set. Assortment intelligence needs complete enumeration of an unknown product set, less often.
That difference drives the collection design: price work optimises for freshness on a fixed list, assortment work optimises for exhaustive traversal. Running both from one pipeline usually means the assortment side is under-served.
8. Summary
Assortment intelligence compares which products retailers actually carry, producing overlap, gap, exclusive, and new-arrival metrics that inform range planning and negotiation.
Product matching across retailers is the hard part and depends on identifiers where they exist and multi-signal inference where they do not; variant handling must be normalised before counts are comparable; and coverage rather than freshness is the collection requirement — so distinguish a genuine absence from a failed crawl before reporting a gap.