Retail Competitive Analysis: What Is Actually Observable

Updated September 2026 · Written and maintained by the Progression Agency strategy team

Most retail competitor analysis fails for one reason: it tries to answer questions the data cannot answer. Nobody outside a retailer knows its margins, its supplier terms or its unit economics. What is observable — assortment, price, availability, promotional cadence, channel mix, review volume and search visibility — is considerably more useful than people expect, and it can be gathered systematically. This page sets out exactly what to collect, how often, and which decisions each field should feed.

What is a retail competitive analysis?

A structured, repeated comparison of your assortment, pricing, availability, promotion, channels and visibility against named competitors, using data that is publicly observable. The essential word is repeated: the value is in the trend, not in the snapshot.

The distinction between this and the generic competitive analysis taught in business courses matters commercially. A generic analysis produces a framework and a conclusion. A retail analysis produces a dataset that tells you on a Tuesday that a competitor has been out of stock in your best category for eleven days, which is a decision you can act on immediately.

Retail competitor analysis and retail competitive analysis are the same thing

Both phrasings appear in practice and refer to the same work. ‘Competitor analysis’ slightly more often describes profiling specific rivals; ‘competitive analysis’ slightly more often describes the category picture. In practical retail use the two are interchangeable.

What can you actually observe about a competitor?

Assortment, price by matched SKU, stock availability, promotional cadence, channel presence, review volume and recency, and organic search visibility. Seven fields, all factual, none requiring any inside knowledge.

Assortment is fully visible and usually under-analyzed

Every product a competitor lists publicly is knowable, which means breadth and depth by category are measurable exactly rather than estimated. Tracked over months, assortment changes reveal category entries and exits well before either is announced.

Price is only meaningful on matched SKUs

Comparing average prices across two ranges is close to meaningless because the ranges differ. Comparing the same product, or a genuinely equivalent one, across retailers over time is the whole value, and building that matched list is the unglamorous work most programs skip.

Availability is the most valuable and least tracked field

Stock status sampled repeatedly turns into a demand signal. A competitor out of stock in a category for two weeks is handing you customers, and the only way to know is to have been checking. This is the field that most often justifies the entire program.

Review velocity is a proxy for sales, not a measurement of it

Review counts grow roughly with volume, so acceleration is informative. It is not a sales figure, the conversion rate from purchase to review varies by category and by retailer, and it should be labeled as a proxy wherever it appears.

What cannot be observed, and should not be guessed?

Margin, supplier terms, unit sales, inventory value, acquisition cost and category profitability. All six are routinely estimated in competitive reports and all six are invented.

This matters more than it sounds. A report containing one number the reader knows to be a guess loses authority for every other number in it, including the ones that were carefully measured. Labeling inference as inference is what makes the measured parts usable.

How do you build the analysis?

Name five competitors, define matched SKUs, choose the observable fields, set a refresh cadence per field, date-stamp everything, and attach a named decision to each field. Then review monthly against your own numbers.

Five competitors, not fifteen

Coverage is the enemy of usefulness here. Five named rivals you genuinely lose sales to produce a dataset you will maintain; fifteen produce one you will abandon in month three. Add a sixth only when you drop one.

Matched SKUs are the foundation

Identify products that are identical or genuinely comparable across every tracked competitor. Fifty matched SKUs is more useful than five thousand unmatched ones, because comparison is only possible where matching has been done.

Every field needs a decision attached

Before adding a field, write down the decision it will inform. Price feeds repricing, availability feeds promotion timing, assortment feeds range planning. A field with no decision attached is why competitor reports stop being read.

The refresh column is the part most often got wrong. Price and availability need weekly or faster sampling to be useful; assortment and visibility change slowly enough that monthly or quarterly is sufficient, and over-collecting them wastes the effort the fast fields need.

What decisions should this actually change?

Repricing, promotion timing, range planning, stock allocation and channel investment. If a competitive analysis has not changed one of those five in a quarter, it is reporting rather than analysis.

The chart sorts findings by how fast you can respond, which is a more useful ordering than how interesting they are. A competitor stock-out is actionable this week; a competitor entering a new category is important and cannot be responded to before next season.

The third row is the trap. Matching a loss-leading discount on a hero product is the most common expensive reflex in retail, and the finding that should trigger it is almost never the price itself but whether the competitor can sustain it.

How do you collect the data?

Manually at first, then partly automated once you know which fields you actually use. Starting with tooling before knowing the fields is how organizations end up paying for data nobody reviews.

Start manual for one month

A spreadsheet, five competitors, fifty matched SKUs and a weekly hour will tell you which fields you genuinely use. Everything you automate after that is automating something proven useful.

Then automate the fast fields only

Price and availability are the fields worth automating, because they need frequent sampling and the collection is repetitive. Assortment and visibility are low-frequency and are frequently better reviewed by a person who notices things a script will not.

Collect within the terms of the sites involved

Observing public pages is ordinary competitive practice; circumventing access controls, creating accounts under false pretences or ignoring stated terms is not. Where a commercial data provider already licenses the data, buying it is usually cheaper than building the collection.

How does search visibility fit in?

It tells you which competitors are capturing demand before it reaches a shelf or a product page. A competitor holding the top positions for your main category terms is intercepting customers earlier in the process than any pricing comparison will show.

Track a fixed list of category terms quarterly and record who ranks. Movement in that list is slow, which is exactly why it is worth watching: a competitor climbing steadily on head terms is building an advantage that will be expensive to reverse once established. Our guide to judging search expertise covers how to assess whether that gap is closeable.

How often should the analysis be reviewed?

Monthly, in a meeting where your own numbers are on the same page. Competitor data reviewed without your own performance beside it produces observations rather than decisions.

What should the output look like?

One page of exceptions rather than a full dashboard. What changed, what it probably means, and what is recommended. Full data stays available for anyone who wants it and does not belong in the monthly review.

Who should own this?

Whoever makes the pricing and range decisions, not whoever collects the data. Analysis owned by a reporting function and consumed by nobody is the standard failure mode, and it is an ownership problem rather than a data problem.

Common mistakes

Seven, and the first two account for most abandoned programs.

For the range decisions this analysis should feed, the product mix guide covers the four dimensions and when to change each. The marketing strategy page connects competitive position to what you say about it.

A starter template: the fields, the cadence and the owner

The table below is the whole program in one view. Copy it, fill in your five competitors, and you have a working specification rather than a project that needs scoping.

The owner column is the one that decides whether this survives. Every row needs a person whose job the decision already is; a field owned by an analyst and consumed by nobody is exactly the program that quietly stops in month three.

Frequently asked questions