Estimating Market Size Without a Research Budget: Top-Down and Bottom-Up

Why market size comes before every growth decision

Many growth plans fail not in execution but because of a number that was never calculated. A business specializes in a niche that is too small, opens a location in a region with too few suitable customers, or plans a service for which there are not a hundred buyers in the catchment area. The analysis specialize or diversify therefore calls market size the most common limit of specialization.

Professional market research is usually too expensive for a small business. So the question is: how does a small business estimate the size of its market with public figures and its own data so that the estimate can support a decision? To answer it, we first clarify the terms and the mental model of decomposition, then the two calculation paths, reconcile them in an example, and close with the errors of judgment that most often distort such estimates.

A rough estimate with open assumptions is worth more than a precise number whose origin nobody knows.

Terms: total market, serviceable market, obtainable market

Three levels help to talk precisely about market size. The total market includes all spending on a type of service, regardless of who provides it and where. The serviceable market is the part a business can in principle serve with its offering, region, and focus. The obtainable market is the part of that which it can realistically win with its sales effort within a manageable period. In English, the three levels are often called TAM, SAM, and SOM.

For decisions, the third level almost always matters. The total market for building services in Germany is large, but it tells a business with twelve technicians in one county little. The obtainable market answers the actual question: how much revenue can this business win with this service here over the next three years?

An estimate is only as good as its assumptions. That is why every number needs its calculation path and a note on where each component comes from: from a statistic, from your own data, or from a justified assumption.

The mental model: breaking large numbers into components

Behind both calculation paths lies the same mental model, decomposition: a quantity that is hard to estimate is split into several easier-to-estimate components, which are then combined. Instead of guessing how much revenue a region yields from maintenance contracts, you estimate how many systems there are, what share is maintained, and what a maintenance visit costs.

McGregor and Lichtenstein (1991) summarize and extend the state of research. In an earlier study, MacGregor, Lichtenstein, and Slovic (1988) had participants estimate quantities such as the number of cigarettes consumed annually in the USA. Compared with direct estimation, decomposition improved the results, especially for very large numbers. In the follow-up study, a further improvement came when components were estimated in several ways and the results combined. But the authors also show the limit: when the information on individual components contained errors, even the best decomposition helped little.

The studies were conducted with lay participants and questions from reference books, not with entrepreneurs and real markets. The mechanism can nevertheless be transferred: a decomposed calculation reveals which assumption carries the result, and that is exactly where you can check in a targeted way.

Calculating top-down: from the total market to your own share

The top-down calculation starts with a published figure, such as industry revenue, the number of businesses, or the number of households, and narrows it down step by step: to your own region, to the matching segment, and finally to a realistic share. Sources include the business register and the statistical series of the Federal Statistical Office (Destatis), the state statistical offices, chambers of commerce and crafts, and industry associations.

The advantage of this calculation is its speed. Its weakness lies in the last step. The share a business can win is almost always set top-down and rarely justified. The well-known formula “If we only get one percent of the market …” is not a calculation but a wish. In addition, published industry definitions rarely match your own service exactly.

Calculating bottom-up: from individual customers to the market

The bottom-up calculation starts with customers. It counts how many possible customers there are in the catchment area, narrows them down to those with a matching need, checks how many your own sales effort can reach, and multiplies the customers realistically won by revenue per customer. This is exactly the niche calculation from the analysis specialize or diversify.

1. Count possible customers

   How many businesses or households in the target group are there in the catchment area? Business directories, chambers, and statistical offices help.

2. Narrow down the need

   What share actually has the need in the coming years? Estimate cautiously and use your own data, such as the share of your existing customers who have it.

3. Check reach

   How many of them can your sales effort approach in three years with the time available? This is often where the real limit lies.

4. Apply a win rate

   What share of those approached will become customers? Use your own close rate from comparable quotes, not a wished-for value.

5. Multiply by revenue per customer

   Apply the average annual revenue per customer from your data and compare the result with your goal.

With the figures from the niche calculation: 1,200 businesses in the target group in the region, 480 of them with a matching need, 240 of those reachable in three years, and at a 15 percent win rate, 36 customers. At 8,000 euros of annual revenue per customer, that is about 290,000 euros. The advantage of the bottom-up calculation is that its assumptions come from your own business: close rate, revenue per customer, and sales time are known or can be measured.

Reconciling both paths: the estimation triangle

The tool of this guide is the estimation triangle. It combines the top-down calculation, the bottom-up calculation, and a third reference point from reality, such as the revenues of known competitors or the number of tenders in a year. If the three values are of the same order of magnitude, the estimate is reliable enough for a decision. If they are far apart, that is not a flaw of the method but its benefit: one of the assumptions is wrong, and the deviation shows where to look.

How to get data on competitors is described in the guide competitor analysis with little effort. For a new region, it is also worth looking at the distance test from expanding into a new region, because distance shrinks the obtainable market more than the map suggests.

Two views: how much precision does a small business need?

The first view considers every market estimate without professional data to be reading tea leaves. The components are uncertain, multiplication amplifies every error, and in the end there is a number that looks more precise than it is. The finding of McGregor and Lichtenstein that faulty component information can cancel the benefit of decomposition supports this view.

The second view counters that the alternative to a rough estimate is not a precise number but none at all. An owner decides anyway whether to enter a niche or region. They do it with or without a calculation. A decomposed estimate at least makes the assumptions visible and testable.

I find the second view more convincing, with one condition: the estimate must be given as a range, not as a point. Calculating with a cautious, a middle, and a favorable assumption shows whether the decision still holds in the cautious case. That is exactly the precision a decision needs.

The most common error of judgment: overconfidence

The greatest danger for a market estimate lies not in the statistics but in the person estimating. In experiments with features of market entry, Camerer and Lovallo (1999) showed that participants who overestimated their own abilities entered markets too often. Their explanation: everyone expects to be better than average and forgets that the others think the same.

It would seem obvious to use reference values as a remedy. Blaseg and Schwienbacher (2024) tested this using the planning data of 971 founders. The result is sobering: after seeing historical success values of comparable projects, the founders distinguished themselves even more strongly from the reference group and afterward overestimated themselves even more. Those who consider their own situation an exception use reference values to justify that exception.

Further biases of this kind are described in the analysis entrepreneurs’ errors of judgment. Why bottom-up planning is more reliable than a goal set from above is argued in the essay realistic growth goals.

Conclusion: precise enough for the decision

How does a small business estimate the size of its market so that the estimate can support a decision? It breaks the question into components, calculates once top-down and once bottom-up, reconciles both with a third reference point, and states the result as a range with open assumptions. Research on decomposing estimation tasks supports this approach as long as the components are right. Research on overconfidence shows where the greatest source of error lies: in the win rate you credit yourself with.

1. This week: Write down the decision for which you need the market size and collect three public figures for the top-down calculation.
2. Next week: Calculate bottom-up with your own close rate and your revenue per customer.
3. In two weeks: Reconcile both results, calculate a cautious variant, and check whether your decision still holds then.

A market estimate does not have to be right to be useful. It only has to say honestly what it is based on.