How to Estimate Market Size for a New Product

A market size estimate is not a number you look up. It is an argument you build from a handful of inputs you can defend, and its value comes from the assumptions being visible — so an investor, a retailer or your own team can challenge one input rather than dismiss the whole figure.

This guide walks through the TAM/SAM/SOM structure, two methods for building the numbers, a worked example, credible free data sources, and the mistakes that make a forecast unusable.

Do both. When top-down and bottom-up land within the same order of magnitude, you have a usable estimate. When they diverge wildly, one of your assumptions is wrong and you have learned something more valuable than the number.

Suppose you are launching a $180 accessory for professional dog groomers in the United States.

Each row can be argued with independently, which is exactly the point. If a reviewer thinks 60% should be 35%, the model updates in seconds instead of collapsing.

Cite the source and the date for every input in a footnote row. An estimate that shows its sources survives due diligence; one that does not gets discounted entirely.

The estimate exists to answer one question: does the obtainable revenue justify the development and tooling investment at your target margin? Run it against three scenarios — conservative, expected and optimistic — and check the conservative case still clears your break-even. If it does not, change the product, the price or the channel before writing engineering requirements.

Market sizing is one input into product discovery. Our product discovery process pairs it with customer interviews and competitive teardown, and our cost guide covers the investment side of the same equation.

Every market size estimate should survive four quick tests. They take an hour and prevent the most common credibility failures in an investor or licensing conversation.

The unit-count test catches more bad forecasts than any other. A $6 million SOM on a $180 product means about 33,000 units a year, roughly 2,800 a month — a real tooling program, a real warehouse and a real support load. If the plan assumed a contract manufacturer and one founder, the number is wrong or the plan is.

TAM is total annual spend if every possible buyer bought. SAM narrows that to buyers you can actually reach with this product, in your channels and regions. SOM is what you can realistically capture in the near term given your sales capacity, budget and competition — it is the only one of the three that should drive planning.

Size the problem, not the product. Count the people or businesses that currently pay to solve it another way, estimate what they spend on that alternative, and model what fraction would switch at your price. Twenty structured customer interviews will validate the switching assumption faster than any secondary research.

Bottom-up is more defensible because every assumption is visible and adjustable. Top-down is a useful cross-check. Build both; if they agree within an order of magnitude, the estimate is usable, and if they do not, find which assumption is wrong before proceeding.

Accurate enough to make a go/no-go decision. If the conservative case comfortably clears break-even, extra precision changes nothing. If the answer flips between the conservative and optimistic cases, that is the signal to spend more on primary research before committing to tooling.

Census business and household counts, Bureau of Labor Statistics occupation counts, trade association shipment data, public company segment reporting, and retailer review and rank data. Combined with customer interviews, these usually produce a better-grounded estimate than a purchased category report.

A market estimate is only useful when it becomes a number of units to build. That translation is where most forecasts break, because the funnel from addressable market to first-year shipments has several stages and each one multiplies down. Doing the arithmetic explicitly prevents the classic mistake of tooling for a fraction of a percent of a large market that never materializes.

Build the first production run against the bottom of this funnel, not the top. Inventory that does not sell is the most expensive mistake available to a hardware business, and a second run ordered because you sold out is a good problem.

Every market size estimate is wrong; the question is whether it is wrong in a way that changes the decision. Before a number goes into a deck, run it through three independent checks. If two of the three land within roughly the same order of magnitude, the estimate is good enough to plan against. If they disagree wildly, the disagreement itself is the finding — usually it means the market definition is doing too much work.

The share plausibility test kills more forecasts than any other. A new entrant taking double-digit share of an established category inside three years is rare enough that it should be argued for explicitly, with a named reason — a structural cost advantage, an exclusive channel, a patent position. Absent that reason, model 1 to 3% and see whether the business still works. If it does not, the problem is the business model, not the estimate.

Write the assumptions on the same slide as the number. A market size estimate presented without its inputs invites a debate about the conclusion; presented with its inputs, it invites a debate about the assumptions, which is the conversation that actually improves the plan.

Keep the model in a spreadsheet where a single cell change re-runs the estimate, and revisit it after the first hundred customers — real conversion and repeat data will beat any secondary source you started with.

Work with LA NPDT: if you are moving from here to execution, start with our product development consulting or talk to us about end-to-end product development.