Size a niche bottom-up: count the businesses or people you could actually sell to, keep only those with the problem, and multiply by what one customer pays per year. Use official business counts, not "global market" reports. For a small app, the useful answer is a count of reachable customers.

This is for founders deciding whether a niche is big enough for a small app or a side business. The running example is a booking app for pet groomers in the United States, using the US Census Bureau's business statistics as of October 2026.

See it in actionNarrow the market until it is real
Narrow 100,000 pet care businesses to 50,000 groomers, 30,000 reachable and 15,000 with the problem. Winning 2% of those is 300 customers. Illustrative example. Use the playback controls or the thin timeline at the top to explore.

In short:

  • Report-seller market sizes describe whole industries, not your slice of one.
  • Government statistics count businesses by industry code, for free.
  • Solo businesses sit in a separate dataset from businesses with staff.
  • Work backwards from your revenue goal to see if the niche is big enough.

Big market reports rarely describe your niche

Search any category and you find press releases like "the pet care market will reach $X billion by 2030". Numbers like these usually include food, vets, insurance and toys. Your booking app sells to none of them.

Top-down numbers also hide how they were made. You can't check the method, so you can't defend the number to yourself, a partner or an investor. A bottom-up count is smaller, but every step is visible.

Bottom-up sizing takes five steps

Each step narrows the count with one filter you can explain.

  1. Count every business or person in the category.
  2. Keep the ones you can reach with your channel and language.
  3. Keep the ones who have the problem, from interviews or reviews.
  4. Multiply by what one customer would pay per year.
  5. Pick a realistic share for the first two or three years.

People often call the levels these steps produce TAM, SAM and SOM: total, serviceable and obtainable market. The total addressable market article on Wikipedia explains the terms. The labels matter less than showing each filter.

Official statistics count businesses for free

Most countries publish how many businesses exist in each industry, grouped by an industry code. In the US, groomers fall under NAICS code 812910, "Pet Care (except Veterinary) Services", which also includes boarding and training.

Two US datasets matter, and you need both:

You can query both on data.census.gov. Developers can also use the County Business Patterns API. Outside the US, the UK's Office for National Statistics publishes UK business counts by activity, size and location. Eurostat publishes structural business statistics for EU countries.

For consumers, the same idea applies with population data. Count the people in the exact group, such as parents of children aged 6 to 12, not "everyone with a phone".

A worked example: groomer bookings in the US

A big category shrinks fast once you apply real filters. The inputs are placeholders to show the method. Replace each one with the real count and your own interview results.

Step Filter Count (example) Source of the real number
All pet care businesses NAICS 812910 100,000 CBP plus Nonemployer
Groomers only About half do grooming 50,000 Interviews, Google Maps sample
Reachable Has a website or Instagram 30,000 Sample 100 by hand
Has the problem Books by phone or text 15,000 Your interviews

At an example price of $20 a month, 15,000 businesses is a $3.6 million yearly ceiling. Winning 2% of them in three years is 300 customers, or about $72,000 a year.

That is a real side business or a small company. It is not a venture-scale market by itself, and the method makes that clear before you build.

Sample by hand where statistics stop

Statistics give you the category. They don't say how many groomers book by text message. For that, take a small sample.

Search Google Maps for groomers in three different cities and open 30 listings. Note which ones offer online booking. If 12 of 30 already do, about 60% don't, which becomes your "has the problem" filter.

Check the count three ways

One method can be off by a factor of ten. Two more quick checks catch that.

  • Search demand. Rough monthly searches for "groomer booking app" in Google Ads Keyword Planner show whether buyers are looking.
  • Competitor claims. Competitors often publish "trusted by 5,000 groomers". If three competitors together claim more customers than your total, your count is too low.
  • Associations and directories. Trade groups often publish member numbers. Compare them with your count.

If the three checks land in the same range, trust the order of magnitude. If they disagree wildly, your category definition is off. That is common, and it's useful to find out now.

Work backwards from your revenue goal

The most useful question is not "how big is the market?" It is "how many customers do I need, and is that a sane share?"

  1. Write your yearly revenue goal, for example $60,000.
  2. Divide by yearly price per customer, $240, to get 250 customers.
  3. Divide by your reachable market with the problem, 15,000, to get 1.7%.

As a rough rule of thumb, a share under about 5% is a plausible target for a focused product. A share over 20% usually means the niche is too small for your goal, or your price is too low. Rethink one of them before you build.

AI agents find sources, you check them

AI research tools are good at finding which datasets and industry codes exist. They are less reliable at reporting the numbers inside them, because a number in an article may be years old or about a different country.

Ask for sources, not answers:

I need to count independent pet groomers in the US.
            List official datasets and industry codes that count them,
            with the URL of each dataset. Do not estimate the count.
            

Then open each dataset yourself and read the number. Our market research workflow with AI agents has a claim-checking table for this step.

From market size to a first version

Once the niche passes the numbers, the next question is whether those customers want your version. In GrowApps, you can describe the app and the customer you sized, and it researches apps like yours before designing screens for your approval (how GrowApps works).

A small published version is often the fastest way to test the "has the problem" filter with real users.

Run your own count in ten minutes

Find your industry code with a search like "NAICS code for [your category]" or your country's equivalent. Open data.census.gov or your national statistics site, look up the count of businesses with that code, and write it down with the year. Then apply one filter you can defend. Next, read how to validate an app idea in a weekend.

Frequently asked questions

What is the difference between TAM, SAM and SOM?

TAM is everyone who could use a product like yours. SAM is the part you can reach with your channel, language and features. SOM is the share you can realistically win in a few years. For a small app, SOM is the number that matters.

What if there is no industry code for my niche?

Start from the closest code, then sample by hand to estimate the share that fits. For consumer apps, start from population data and narrow by age, location or behaviour.

Sources and further reading