How to Size a Market Using Consumer Data

Every investor deck has a market size slide. Almost none of them are reliable. The typical approach is to cite an industry report claiming the global widget market will be worth $47 billion by 2028, divide by some arbitrary fraction to arrive at your “addressable” share, and present a number large enough to justify investment. This is not market sizing. It is market theater. Real market sizing starts from consumer behavior and works upward, not from analyst forecasts and works downward.

Why Top-Down Sizing Is Unreliable

Industry analysts produce market size estimates by surveying vendors, aggregating revenue data, and extrapolating growth curves. The resulting numbers describe a broad industry, not your specific opportunity. When a report says the meal kit market is worth $4 billion, that includes every meal kit sold by every company to every customer through every channel. It tells you almost nothing about how many consumers would buy yourmeal kit at your price through yourdistribution model.

The TAM/SAM/SOM framework is meant to narrow this. In practice, the narrowing factors are arbitrary. Teams apply percentage discounts based on geography, demographics, or competitive share, but these are guesses. A team that estimates its SOM as 2% of SAM has chosen a number that feels plausible. There is no methodological basis for 2% rather than 0.5% or 5%. The precision is false, and the estimates are often off by an order of magnitude.

Bottom-Up Sizing From Consumer Behavior

Bottom-up sizing works differently. Instead of starting with the total market and shrinking it, you start with the individual consumer and scale up. The inputs are: how many consumers match your target profile, what percentage buy in your category, how often they buy, and how much they spend per transaction. Multiply these together and you have an estimate grounded in observable behavior rather than analyst projection.

Suppose you are launching a premium pet food brand. Bottom-up sizing starts with the number of households with dogs in your target geography, narrows to those currently buying premium (which purchase data can estimate), applies average monthly spend in that tier, and calculates annual category spend for the segment. Every input corresponds to actual consumer spending, not a theoretical maximum derived from industry totals.

The key advantage: every input is individually testable. With real purchase behavior data, you can check how many people match your target profile, measure their current spending, and observe their purchase frequency. Each assumption can be validated, which means the overall estimate is as strong as its weakest assumption rather than as weak as its broadest one.

The Two Inputs Teams Get Wrong

Purchase frequency and average spend are the most important inputs to bottom-up sizing, and both are routinely misjudged by teams that rely on category averages.

Purchase frequency varies enormously within categories. In coffee, some consumers buy beans weekly; others buy monthly. In skincare, some repurchase every six weeks; others buy once a year. Using category-average frequency produces a meaningless aggregate. The useful approach is to segment by frequency and size each tier separately. Heavy buyers and light buyers represent fundamentally different revenue opportunities, and sizing them together obscures both.

Average spend is similarly misleading when aggregated. If your target consumer spends $30 per purchase but the category average is $15 (dragged down by budget buyers you are not targeting), the category average will halve your estimate. The spend figure must match your target segment, not the market as a whole. Purchase behavior data lets you isolate spending within your specific price tier rather than relying on blended averages that mix your target with consumers you will never reach.

Defining the Segment Behaviorally

The hardest part of bottom-up sizing is defining who your target consumer actually is. Your addressable segment is not “women aged 25 to 45.” It is consumers who currently buy in your category, at your price tier, through channels you can reach, with sufficient frequency to represent a viable customer.

Purchase data and behavioral segmentation let you define this segment precisely and, more importantly, count it. You can identify consumers who buy competing products, measure their spending patterns, and estimate the population that matches. This is a fundamentally different exercise from drawing demographic boundaries and guessing what percentage falls inside them.

The segment definition also reveals whether your market supports your business model. If bottom-up sizing shows that only 50,000 consumers match your behavioral profile and each would spend $200 per year, your maximum addressable revenue is $10 million. That might be a perfectly good business, or it might be too small for your ambitions. Either way, knowing this before you commit resources is better than discovering it after.

What Purchase Data Can and Cannot Validate

Consumer research grounded in purchase behavior can validate several critical sizing assumptions. Does your target segment exist in the size you believe? Are they dissatisfied enough with current options to consider switching? Is your price point within the range they currently pay? Is the purchase frequency you assume consistent with observed behavior?

Each of these can be tested before you build anything. Run a concept test against consumers who match your target behavioral profile. Measure purchase intent and price sensitivity. If your model assumes 30% intent among the target segment but research shows 12%, your estimate is roughly 2.5 times too high.

Be honest about the limits. Purchase intent from a synthetic panel, even one grounded in behavioral data, is not a conversion rate. It is a directional signal. Bottom-up sizing built on behavioral data is more reliable than top-down sizing built on analyst reports, but it is still an estimate with uncertainty around it. The advantage is that the uncertainty is bounded and each component can be stress-tested independently. With top-down sizing, the uncertainty is hidden inside a single large number that no one can decompose.

Sizing Should Evolve

Most teams treat market sizing as a one-time exercise: build the slide, get the funding, move on. Your pre-launch sizing assumptions are hypotheses. Post-launch data validates or invalidates them. Actual customer acquisition costs, conversion rates, retention, and average revenue will differ from your estimates, and your market size figure should update as real data replaces projections.

A market size figure built bottom-up from behavioral data will be less impressive on a slide than one built top-down from analyst reports, but especially when researching an unfamiliar market, it will be far more useful for deciding where to invest, how to price, and how aggressively to grow. The point of sizing a market is not to produce the largest defensible number. It is to produce the most accurate one.