Segmentation means dividing a market into parts that behave differently. A report might split a market by what is sold, who buys it, what it is used for or how it reaches the buyer. Regional breakdowns are a particular kind of segmentation based on geography. Both turn a headline total into something you can act on, and both are easy to get wrong.
Common ways to segment a market
| Dimension | Splits by | Useful when |
|---|---|---|
| Product or type | Variants of what is sold, such as material, format or technology | Products differ in price, margin or growth |
| Application or use | What buyers use the product for | The same product serves very different needs |
| End user or customer | Industry, household type or organisation size | Buying processes differ by customer |
| Channel | Direct sales, distributors, retail, online | Route to market affects pricing and reach |
| Price tier | Economy, mid-range, premium | Growth is concentrated at one end of the market |
| Geography | Regions, countries, sometimes cities | Regulation, income or infrastructure varies by place |
Two rules every segmentation should follow
Mutually exclusive
Each unit of the market should belong to one segment only within a given dimension. If a product can be counted under both "industrial" and "commercial" applications, the segments overlap and their values cannot be added together. Reports that let segments overlap often produce totals larger than the market itself.
Collectively exhaustive
Together the segments should cover the whole market. This is why many tables include an "other" row. A large "other" category is worth noticing: if it is a quarter of the market, the segmentation is not explaining much.
Reading regional breakdowns
Regional tables usually follow a fixed hierarchy: a handful of world regions, each broken into a few major countries and a "rest of region" line. That structure is convenient, but it does not always fit the market. Ask these questions:
- Do the regions sum to the global total? Check the same year in each table.
- Is the regional split based on where products are made, sold or used? These can differ a lot for traded goods.
- Is the country list specific to this market? A list identical to every other report from the same publisher suggests a template rather than analysis.
- Are the "rest of" lines plausible? These residual figures are often the least researched.
- Is currency conversion explained? Exchange-rate movements can create or erase apparent growth in regional values.
- Does commentary differ by region? Real regional differences should produce different explanations.
A simple consistency test
The figures below are hypothetical and illustrate the arithmetic only.
Imagine a report gives a global total of 500 and three regional values of 210, 180 and 140. Those sum to 530, which is 30 more than the total. Either the regions overlap, the total was calculated on a different basis, or one of the figures is wrong. A gap of a few units can be rounding; a gap of six percent needs an explanation. Run the same test on product segments, and across years: shares that jump around without explanation suggest the series was not built from a single consistent model.
Segment growth rates
Reports often highlight the "fastest-growing segment". Two cautions apply. First, small segments can post high percentage growth from a low base, so look at absolute change as well. Second, if every segment grows faster than the total, the arithmetic is broken: a total's growth rate is a weighted average of its parts, so at least one part must grow at or below it. You can use our CAGR explainer to recompute segment rates from the tables.
Matching segments to your question
The best segmentation for a report's publisher is not always the best one for you. A publisher may segment by product type because that is how supply-side data is available, while your decision depends on customer size. When segments do not match your question, the report may still help with the total, but you will need other evidence for the split that matters. Knowing how the publisher built the split, covered in our research methods explainer, tells you how far to trust it.
Unfamiliar term? See the glossary.