Amazon category analysis: build an evidence-led category report

By Scoutvo · Published October 16, 2026 · Updated October 16, 2026 · 5 min read

Define a decision before collecting a category

Amazon category analysis compares a defined set of products to answer a commercial question. Start with a customer need and product constraints: for example, which desk chairs offer comparable adjustment functions in the delivery context you serve? A browse category alone does not guarantee substitutes. A gaming chair, a task chair and a stool can share search results while solving different problems.

This guide owns the category-level workflow. For selecting individual rivals, use finding Amazon competitors; for a listing-by-listing audit, use the competitor listing worksheet. Keep your scope narrow enough that a reviewer can explain every inclusion and exclusion.

1. Make the sample reproducible

Record marketplace, query or browse node, collection time and timezone, delivery context, language and currency. Save ASIN, child and parent identifiers where known, selection route and exclusion reason. Distinguish organic results, sponsored placements and manually added products. Search results can vary; save the actual selection, rather than assuming another search will reproduce it.

Decide the observation unit before calculating. One row per child may weight a large variation family heavily. One row per family removes that weighting but hides child-level differences. Neither choice is automatically correct. Report the chosen unit and disclose uncertain family assignments. Keep your own product separate when describing the competitor distribution.

2. Separate observation, estimate and unknown

Store a reason for missing data. Failed retrieval, an unavailable offer and a genuinely absent field are different states. A blank price is never a zero-price product. Any modeled sales column remains an estimate with a named model and uncertainty; it does not become a measured sale because it appears beside observed fields.

3. Worked export: what can actually be calculated?

An authorized Scoutvo office-chair export labeled August 24, 2026 contains 71 unique ASIN rows: one own product and 70 peer rows. All rows use EUR. The export does not contain a marketplace or observation timestamps. The filename label is an export label, not a verified collection date. This is a limited export audit, not a current Amazon marketplace study.

The median uses the mean of the 24th and 25th sorted valid values; the unrounded result is €89.455. Parsing uses the export's decimal comma. The own-product row is excluded. No shipping, pack-size or functional normalization was possible from this calculation. Parent-family weighting and the original selection process are unverified. The 22 missing observations could materially change the distribution, so the median is not a category-wide reference price.

The defensible result is that this export can support a raw-price summary with an explicit denominator. It cannot establish demand, a profitable selling price, market share, or a price-related cause of rank. The machine-readable aggregate and limitations make the calculation auditable without publishing seller identifiers or the raw export.

4. Improve comparability before interpreting position

For a fresh study, group products by verified use, size, pack and important functions before comparing total customer prices. Check a small subset manually against the current detail pages. Record conflicting specifications and unresolved fields rather than silently treating a title claim as a test result.

Calculate summaries for each defensible group with valid-row counts. Do not turn a subgroup of two products into a robust market benchmark. Compare a refreshed collection against a fixed sample and separately record additions and removals. Otherwise a changing selection can look like a changing market.

5. Convert findings into questions you can test

A common review complaint can prompt a product investigation. A supported specification missing from your listing can prompt a copy change. A different price position can prompt a check of pack, shipping and costs. None of these observations proves an unmet market or a profitable launch.

Use review sampling and aspect coding to validate customer-language themes, and consult authorized account or query reports for demand evidence. Assign each proposed action an owner, evidence source, verification step and review date. Keep commercial assumptions visibly separate from collected facts.

6. Retain the method with the report

A useful handover contains the sample manifest, exclusions, raw observations, cleaning rules, denominators and unresolved questions. Add the dates and marketplace rather than relying on a screenshot label. The public office-chair example deliberately records those two fields as unknown because the source does not establish them.

Use the local price calculator and checklists for a first review. Open the Scoutvo sample report to inspect presentation, then ask which fields, history and sources are available for your actual category. A clean report explains the boundary of its evidence as clearly as its findings.