> Enrich.chat gives AI agents verified work emails and mobiles. To use it from an agent, read https://enrich.chat/agents.md . This is the Markdown version of https://enrich.chat/blog/tam-sam-som/

# TAM, SAM and SOM: how to calculate them from company counts

TAM is every possible buyer, SAM is the part you can serve, and SOM is the part you can win. Here is the bottom-up method, with a worked example built on public Census firm counts.

By Enrich.chat Editorial. Updated October 16, 2026.

**Short answer:** TAM is the revenue if every possible buyer bought from you, SAM is the part you can serve, and SOM is the part you can win soon. To size them bottom-up, count companies by industry and size, then multiply by price. The [Census Bureau's firm counts](https://www.census.gov/programs-surveys/susb/about.html) are a free source for US numbers.

## Key takeaways

- TAM, SAM and SOM are nested: each is a smaller slice of the one before, and only SOM belongs in a revenue plan, per HubSpot.
- Bottom-up sizing is companies counted times price per company. Top-down sizing multiplies a market report by guessed percentages, and in our example one guess moves the answer fourfold.
- The 2022 Census firm table lists 157,072 US truck transportation firms with paid employees, and only 794 of them have 200 to 999 employees.
- Those 794 firms are 0.5% of firms but 11.2% of receipts, so price each size band separately instead of using one average.
- Enrich.chat counts matching companies in its own records for free, which is not a census of the market. It also checks accounts you already hold: 5 credits for each company profile found, and 0 when nothing is found.

## What are TAM, SAM and SOM?

TAM, SAM and SOM are three nested estimates of how much revenue a market holds. TAM (total addressable market) is the revenue you would earn if every possible buyer bought from you. SAM is the part of it that your product, region and customer profile can serve. SOM is the part you can realistically win in a set time. [HubSpot](https://blog.hubspot.com/marketing/tam-sam-som) describes TAM as the maximum potential demand of a market, SAM as the part of TAM you can reasonably target, and SOM as the part of SAM you can potentially convert.

| Term | Plain meaning | Question it answers | Use it for |
| --- | --- | --- | --- |
| TAM | Every company that could buy, times what each would pay | Is the prize big enough to chase? | Context and pitch decks |
| SAM | The companies your product, region and ICP can serve | Who can we sell to today? | Targeting and territory plans |
| SOM | The part of SAM you can win in a set period | What can we plan to sell? | Revenue forecasts and budgets |

Guides disagree on one word. HubSpot and [Carta](https://carta.com/blog/market-size/) spell out SAM as "serviceable addressable market". [Salesforce](https://www.salesforce.com/blog/small-business/tam-sam-som/) and [Wikipedia](https://en.wikipedia.org/wiki/Total_addressable_market) write "serviceable available market". The arithmetic is the same. Only SOM belongs in a revenue plan, because HubSpot warns that treating TAM as achievable revenue leads to unrealistic projections.

### How do investors read the three numbers?

Investors ask about market size early. Carta says venture capitalists have no fixed cut-off, but a good rule of thumb for a venture-backed startup is a TAM over $1 billion. It describes SOM as typically your three to five year revenue projection. A smaller TAM is not a flaw if you do not need venture funding, and the worked example below has a TAM under $1 billion on purpose.

## How do top-down and bottom-up market sizing differ?

Top-down sizing starts with a published market figure and cuts it with percentages. Bottom-up sizing counts the companies and multiplies by what each would pay. HubSpot calls top-down quick but dependent on assumptions, and bottom-up more defensible but slower. Salesforce draws the same line and says top-down suits a new product with little sales data.

```
Top-down:   market report x share you can serve x share you can win
Bottom-up:  companies counted x price per company
```

The weak point of top-down sizing is that every percentage is a guess. In a 2014 essay on market size, investor [Bill Gurley](https://abovethecrowd.com/2014/07/11/how-to-miss-by-a-mile-an-alternative-look-at-ubers-potential-market-size/) argued that two core assumptions in a widely cited valuation, the size of the existing taxi market and the share a newcomer could win, could be off by a factor of 25 or more, because a better product can create new demand. The weak point of bottom-up sizing is the count, since you need a defensible number of companies. The next section covers where to get one.

Use both methods. HubSpot says most businesses benefit from running both, and that a large gap between them is a signal to investigate before you present either number.

## Where do you get company counts for a bottom-up estimate?

For US companies, the Census Bureau's [Statistics of U.S. Businesses (SUSB)](https://www.census.gov/programs-surveys/susb/about.html) gives free counts of firms by industry, state and company size. Census says private businesses use it for market research and for managing sales territories. As of October 16, 2026, the newest year on the [SUSB tables page](https://www.census.gov/data/tables/2022/econ/susb/2022-susb-annual.html) is 2022. Know four limits before you use it:

- **It counts employers only.** The [SUSB glossary](https://www.census.gov/programs-surveys/susb/about/glossary.html) says the series leaves out businesses with no paid employees, which sit in a separate series, [Nonemployer Statistics](https://www.census.gov/programs-surveys/nonemployer-statistics.html), whose latest year is 2024. The gap can be wide. The UK's [business population estimates](https://www.gov.uk/government/statistics/business-population-estimates-2025/business-population-estimates-for-the-uk-and-regions-2025-statistical-release) count about 5.7 million private-sector businesses at the start of 2025, of which about 4.3 million had no employees and about 1.4 million were employers. Decide whether one-person businesses can buy from you.
- **It counts firms, not places.** [County Business Patterns](https://www.census.gov/programs-surveys/cbp.html), whose latest year is 2023, counts establishments, meaning locations. A carrier with ten depots is one firm and ten establishments. For account counts, use firms.
- **Size means the whole company.** The glossary defines an enterprise as all the establishments under common ownership, and the size bands use its total employment. A 300-person firm with one trucking depot lands in the trucking row, in the 300 to 399 band.
- **It runs two years or more behind.** Census says SUSB data arrive about two to two and a half years after the reference year, and the 2022 tables carry small random noise added for privacy, so counts are close rather than exact. HubSpot advises using data from the last two years, which SUSB cannot meet, so write the year beside every count.

If you would rather not open spreadsheets, [Census Business Builder](https://www.census.gov/data/data-tools/cbb.html) is a free Census tool with maps and downloadable reports on businesses in a chosen place and industry. Industry rows use NAICS codes, and the 2022 SUSB tables use the 2017 version. Our guide to [firmographic data](https://enrich.chat/blog/firmographic-data/) explains how the codes work and change.

### When do public counts fall short?

Public statistics stop at industry, place and size. They do not list company names or websites, and they cannot filter by the software a company runs, which is the job of [technographic data](https://enrich.chat/blog/technographic-data/). Outside the US you need each country's own tables. When you need names, websites or technology filters, you need a B2B database or data provider. Our guides to the [B2B marketing database](https://enrich.chat/blog/b2b-marketing-database/) and the [best company data APIs](https://enrich.chat/blog/best-company-data-apis/) compare the options. Enrich.chat's company search is one such option, with a free count of matches.

## How do you calculate TAM, SAM and SOM from a company count?

Count the companies in each size band, price each band, add the bands for TAM, keep the bands that match your ICP for SAM, and apply your capacity and win rate for SOM. The example sizes the market for route-planning software sold to US freight carriers, the same fictional product as in our [ideal customer profile guide](https://enrich.chat/blog/ideal-customer-profile/). The firm counts are real Census figures. The prices, capacity and win rate are made up, so replace them with yours.

1. **Define the market in one line.** Here it is US firms in truck transportation, NAICS 484.
2. **Pull the counts.** In the [2022 SUSB table](https://www.census.gov/data/tables/2022/econ/susb/2022-susb-annual.html) for the US by NAICS and detailed employment size, the NAICS 484 rows list firms in every size class. The total is 157,072 firms with paid employees.
3. **Group the size classes into bands you can price.** The table below uses four.
4. **Price each band.** A 10-person firm and a 2,000-person firm will not pay the same.
5. **Add the bands.** The total is TAM.
6. **Filter to your ICP for SAM.** The ICP here is 200 to 999 employees. The ICP guide says 200 to 1,000, but the Census size classes break at 999 and 1,000.
7. **Apply capacity and win rate for SOM.** The SOM section below shows how.

| Size band (employees) | US firms, 2022 (Census) | Price a year (assumed) | Market a year |
| --- | --- | --- | --- |
| Under 20 | 145,826 | $1,200 | $175.0 million |
| 20 to 199 | 9,986 | $6,000 | $59.9 million |
| 200 to 999 (the ICP) | 794 | $30,000 | $23.8 million |
| 1,000 or more | 466 | $80,000 | $37.3 million |
| Total (TAM) | 157,072 |  | $296.0 million |

The firms are counted by Census and the prices are assumptions. That makes TAM about $296 million and SAM about $23.8 million, the 794 firms in the ICP band at $30,000 each. Copy the table into a spreadsheet, keep the Census counts and replace the price column with yours. The market column is the count times the price.

### Why price each size band separately?

Count and money tell different stories. Firms with under 20 employees are 92.8% of the 157,072 firms but account for 26.7% of the $451.3 billion in receipts that SUSB reports for the industry. The 794 firms with 200 to 999 employees are 0.5% of firms and 11.2% of receipts. One average price across all firms would overstate the small end or understate the large end, and either error flows straight into TAM.

**Figure 1: US truck transportation firms and receipts by company size, 2022.** Firms with under 20 employees are 92.8% of 157,072 US truck transportation firms but 26.7% of receipts, while the 794 firms with 200 to 999 employees are 0.5% of firms and 11.2% of receipts, and the 466 firms with 1,000 or more employees are 0.3% of firms and 37.9% of receipts.

*Share of firms and share of receipts by company size, NAICS 484, 2022. The 0.5% and 0.3% firm bars are drawn at a minimum width so you can see them. Source: [Census Bureau, SUSB 2022 table](https://www.census.gov/data/tables/2022/econ/susb/2022-susb-annual.html).*

### How do you cross-check with a top-down number?

Take industry receipts and apply a guessed share. SUSB reports $451.3 billion in receipts for NAICS 484 in 2022, and the glossary notes that summed receipts duplicate some revenue for large multi-establishment firms. If you guess that carriers spend 0.1% of receipts on route-planning software, top-down TAM is $451 million.

| Guessed share of receipts | Top-down TAM |
| --- | --- |
| 0.05% | $226 million |
| 0.10% | $451 million |
| 0.20% | $903 million |
| Bottom-up result (about 0.07%) | $296 million |

Moving one guess from 0.05% to 0.2% changes the top-down answer fourfold, while the bottom-up number moves only when a count or a price changes. The bottom-up TAM implies carriers spend about 0.07% of their receipts on this kind of software. That gives you a concrete question for customer interviews. If the two methods land far apart, HubSpot says to find out why before presenting either.

## How do you set a realistic SOM?

Base SOM on the accounts you can work and the share you win, not on a slice of the whole market. HubSpot divides last year's revenue by last year's SAM to get your market share, then applies that share to this year's SAM, which needs sales history. For a new product, Salesforce's bottom-up example instead multiplies the customers its sales team can win in a year by the price.

In the worked example, two reps each work 160 accounts, so 320 of the 794 SAM firms get attention in year one. At a 10% win rate that is 32 customers. At $30,000 each, SOM is $960,000, or 4.0% of SAM. Replace both inputs with numbers from your own pipeline as soon as you have one. A SOM built this way also tells you how many accounts to research, which is where account-level data comes in. Our guide to [account-based marketing](https://enrich.chat/blog/account-based-marketing/) covers how to work a named list. Figure 2 puts the three numbers side by side.

**Figure 2: TAM, SAM and SOM for a US trucking software example.** Three nested circles show a TAM of about $296 million from 157,072 US truck transportation firms, a SAM of about $23.8 million from 794 firms with 200 to 999 employees, and a SOM of about $0.96 million from 32 customers.

*TAM, SAM and SOM for the example. Firm counts come from the [Census Bureau's 2022 SUSB table](https://www.census.gov/data/tables/2022/econ/susb/2022-susb-annual.html) for NAICS 484. The prices, accounts worked and win rate are made up. The circles are not drawn to scale.*

## What mistakes make a market size wrong?

The most common mistakes are unsourced counts, one price for every company, and mixing definitions of a company. Watch for these:

- **Using a count you did not look up.** Round numbers travel from guide to guide. The bottom-up example in [Salesforce's guide](https://www.salesforce.com/blog/small-business/tam-sam-som/) imagines about 600,000 US manufacturers, 200,000 of them with 50 to 100 employees. It is a made-up case, but compare it with SUSB 2022: 239,265 manufacturing firms with paid employees, 13,324 of them with 50 to 99 employees, and 133,388 firms of that size across all industries together. Firms with 50 to 99 employees are employers, so leaving out one-person businesses cannot explain the gap.
- **One price for every size.** Figure 1 shows why. A single average hides where the money is.
- **Mixing firms, locations and one-person businesses.** The same industry has 157,072 firms and 174,556 establishments. Either count is right, and a total that mixes them is wrong. In the UK data, counting non-employers raises the total about fourfold.
- **Leaving the year off.** A count with no date cannot be checked. SUSB 2022 is the newest firm table by size, so say so, and review the estimate at least once a year, as HubSpot advises.
- **Treating TAM as a forecast.** TAM is a ceiling. Put SOM in the plan.
- **Skipping the cross-check.** Run the top-down and bottom-up numbers side by side and explain the gap.
- **Comparing different code versions.** The 2022 SUSB tables use 2017 NAICS codes. Codes can change between versions, so compare like with like.

## How does Enrich.chat fit into market sizing, and where does it stop?

Enrich.chat does not size a market for you, but it can count and check companies. `GET /v1/companies/count` is free and returns how many companies in its records match filters such as industry, state, size band, revenue band and technology used. That counts what Enrich.chat holds, not every company in a market, so use it beside the Census tables for step one, not in place of them. It can also check named accounts you already hold: look up a company by website for its industry, size band and revenue band, test whether you can find a buyer there, and return a work email with a status. It is contact and company data for AI agents and apps. An agent reads [enrich.chat/agents.md](https://enrich.chat/agents.md) or adds the MCP server at `https://enrich.chat/mcp`, and the [install page](https://enrich.chat/install/) has the steps.

### What can an agent do with a list of company websites?

| Job | Endpoint | Returns | Credits |
| --- | --- | --- | --- |
| Count matching companies | `/v1/companies/count` with filters | How many companies in its records match | 0, it is free |
| Classify an account | `/v1/company` with a domain | Name, industry, size band, revenue band, state, country | 5 when found |
| Check you can reach a buyer | `/v1/company/people` with a domain, title and seniority | Name, title, LinkedIn URL | 5 for each person returned |
| Sample buyers without a list | `/v1/search` with a plain-English query and filters | Up to 25 people a page | 5 for each person returned |
| Get a work email | `/v1/person` with a LinkedIn URL | Work email with a status | 10 when found |

The industry on a company profile is a plain-word label such as logistics and supply chain. It is not a NAICS code, and a label can be broader or narrower than NAICS 484, so you or your agent decide which labels count as trucking.

### What does a lookup return?

Here is an illustrative request and response for a fictional company. The size band and state come straight from the profile.

```
GET https://enrich.chat/v1/company
  ?domain=northlinefreight.example.com

{
  "found": true,
  "id": "org_2Lm8xKq4WzP7RnYt",
  "name": "Northline Freight",
  "domain": "northlinefreight.example.com",
  "industry": "logistics & supply chain",
  "size": "201-500",
  "employees_on_linkedin": 420,
  "revenue": "50M-100M",
  "founded": 2009,
  "city": "Dallas",
  "state": "Texas",
  "country": "United States",
  "credits_used": 5,
  "free_lookups_remaining": 9
}
```

The size band 201-500 sits inside the 200 to 999 SAM, so Northline Freight is in if you decide its industry label means trucking. A band of 501-1000 straddles the 999 and 1,000 break, so check those by hand. `employees_on_linkedin` is a signal for size, not an exact headcount. A lookup that finds nothing returns `{"found": false}` and uses 0 credits. The next request checks whether a buyer can be found at an account that fits.

```
GET https://enrich.chat/v1/company/people
  ?domain=northlinefreight.example.com
  &seniority=vp
  &title=operations
  &limit=1

{
  "found": true,
  "count": 1,
  "domain": "northlinefreight.example.com",
  "people": [
    {
      "id": "per_7Qd3vNb5XsK2WjLm",
      "name": "Priya Raman",
      "title": "VP Operations",
      "company": "Northline Freight",
      "level": "vp",
      "has_email": true,
      "has_mobile": false,
      "linkedin_url": "https://www.linkedin.com/in/priya-raman-example"
    }
  ],
  "total_matches": 3,
  "next_cursor": "eyJyIjoxfQ",
  "credits_used": 5,
  "free_lookups_remaining": 8
}
```

On the free tier, `free_lookups_remaining` falls by one for each lookup. Passing the LinkedIn URL to `/v1/person` returns the work email and its status, one of `verified`, `catch_all` or `unknown`. A LinkedIn URL is only a lookup key, and Enrich.chat does not log in to LinkedIn. Treat every contact as a strong lead, not a guarantee.

**Figure 3: Which market sizing step needs a database and which can use Enrich.chat.** Four steps of a bottom-up estimate. Counting the whole market needs a Census table or a database. Enrich.chat can count only the companies in its own records. Classifying named accounts uses the company endpoint, checking for a buyer uses the company people endpoint, and getting a work email uses the person endpoint.

*Which sizing step needs a database and which can use Enrich.chat. Credits are spent only on matches, and a lookup that finds nothing uses 0. Endpoints come from [Enrich.chat's agents.md](https://enrich.chat/agents.md).*

### How much does the sample cost?

Suppose you hold 300 company websites from your CRM or a trade directory. Classifying all of them uses 1,500 credits at 5 each, if every one is found. Checking for one VP of Operations at 40 accounts that fit the SAM uses 200 more, and a work email for each of those 40 people uses 400 more. The whole job is 2,100 credits, which fits the Growth plan at $29 a month. That is about 380 requests, at about one request a second, so roughly six minutes. To skip the code, upload the 300 websites to the dashboard's [Bulk enrich](https://enrich.chat/features/bulk-enrichment/), which enriches companies from a CSV. Choose "Run it on the server" and the [background job](https://enrich.chat/features/background-jobs/) keeps going after you close the tab. You also get 10 free API lookups a day with no signup, enough to test the method on ten companies before you pay. See [pricing](https://enrich.chat/pricing/) for every plan.

```text
Use www.enrich.chat to check my list of company websites against this SAM: US trucking companies with 200 to 999 employees. For each website, look up the company and return its industry, size band and state, then label it in or out of the SAM with one reason. For the first 40 that are in, find one VP of Operations. Mark any website with no match as not found, and report the total credits used.
```

### What are the limits?

- **It counts its own records, not the market.** The free company count covers the companies Enrich.chat holds, so use Census tables for the full count. A server job holds at most 5,000 rows and keeps its results for 30 days, so download the file when it finishes. The free [lead list builder](https://enrich.chat/tools/lead-list-builder/) runs one search at a time.
- **Companies must be websites.** A company lookup takes a domain, a LinkedIn company page or a company ID, and company names alone are not supported. A batch call takes up to 25 per call, at about one call a second.
- **The company profile has limits.** It has name, domain, industry, size band, revenue band, founded year, location, latest funding stage, growth and web traffic where it has them. Revenue is a band, not an exact figure, and there is no parent company, so your price per customer has to come from your own deals.
- **Search returns 25 people a page.** A cursor pages on up to 1,000. A request that is too broad, such as a role alone, comes back with a hint instead of results.

For account-level work, see the [company enrichment API](https://enrich.chat/api/company-enrichment/) and the [people search API](https://enrich.chat/api/people-search/). Our guides to [company data enrichment](https://enrich.chat/blog/company-data-enrichment/), [finding decision makers](https://enrich.chat/blog/decision-makers-in-a-company/) and [B2B lead generation strategies](https://enrich.chat/blog/b2b-lead-generation-strategies/) cover what to do with the accounts afterwards. If you contact the people you find, you are responsible for rules such as CAN-SPAM, TCPA and GDPR. Anyone can ask for removal at [enrich.chat/opt-out/](https://enrich.chat/opt-out/), processed within 15 days, and accounts are covered by our [terms](https://enrich.chat/terms/). This is not legal advice.

## Frequently asked questions

### What is the difference between TAM, SAM and SOM?

TAM is the revenue if every possible buyer bought from you. SAM is the part of TAM your product, region and customer profile can serve. SOM is the part of SAM you can realistically win in a set period, such as the next year. Each is smaller than the last, and [HubSpot](https://blog.hubspot.com/marketing/tam-sam-som) advises using SOM, not TAM, for financial projections.

### How do you calculate TAM, SAM and SOM?

Bottom-up, TAM is the number of companies in the market times the price each would pay in a year. SAM is the same sum for only the companies that fit your product and ideal customer profile. SOM is the accounts you can work times your win rate times the price. For US firm counts by industry and size, use the [Census Bureau's SUSB tables](https://www.census.gov/data/tables/2022/econ/susb/2022-susb-annual.html).

### Is top-down or bottom-up better for market sizing?

Bottom-up is easier to defend, because every input is a count or a price you can check. Top-down is faster but rests on percentages that are guesses. HubSpot recommends using both and investigating when they land far apart. In our worked example, one guessed share moves the top-down answer fourfold while the bottom-up number stays put.

### Where can you find the number of companies in an industry?

In the US, the Census Bureau's Statistics of U.S. Businesses gives firms by industry, state and employee size, but it covers employers only and runs about two years behind. The UK publishes business population estimates. For company names, websites or technology filters you need a B2B database, covered in our [B2B marketing database guide](https://enrich.chat/blog/b2b-marketing-database/).

### What percentage of SAM is a realistic SOM?

No standard percentage exists. HubSpot derives SOM from your share of SAM last year, and Salesforce multiplies the customers a sales team can win in a year by the price. Build SOM from the accounts you can work times your win rate. In our example, 32 wins from 794 accounts is 4.0% of SAM, using made-up inputs you should replace.

### Can Enrich.chat calculate my TAM?

Not on its own. Enrich.chat's company search shows a free count of the companies in its records that match your filters, but that counts what it holds, not the whole market, and it does not price anything. It can also look up named companies by website for industry, size band and revenue band, find a decision maker, and return a work email with a status. See the [market sizing use case](https://enrich.chat/use-cases/market-sizing/) and the [company enrichment API](https://enrich.chat/api/company-enrichment/) for details.

## Sources

1. [TAM, SAM & SOM: What Do They Mean & How Do You Calculate Them?](https://blog.hubspot.com/marketing/tam-sam-som)
2. [TAM, SAM, and SOM: Made Simple for Growing Businesses](https://www.salesforce.com/blog/small-business/tam-sam-som/)
3. [Market Size: What is TAM, SAM, & SOM?](https://carta.com/blog/market-size/)
4. [Total addressable market](https://en.wikipedia.org/wiki/Total_addressable_market)
5. [How to Miss By a Mile: An Alternative Look at Uber's Potential Market Size](https://abovethecrowd.com/2014/07/11/how-to-miss-by-a-mile-an-alternative-look-at-ubers-potential-market-size/)
6. [2022 SUSB Annual Data Tables by Establishment Industry](https://www.census.gov/data/tables/2022/econ/susb/2022-susb-annual.html)
7. [About Statistics of U.S. Businesses (SUSB)](https://www.census.gov/programs-surveys/susb/about.html)
8. [Statistics of U.S. Businesses (SUSB) Glossary](https://www.census.gov/programs-surveys/susb/about/glossary.html)
9. [County Business Patterns (CBP)](https://www.census.gov/programs-surveys/cbp.html)
10. [Nonemployer Statistics (NES)](https://www.census.gov/programs-surveys/nonemployer-statistics.html)
11. [Census Business Builder (CBB)](https://www.census.gov/data/data-tools/cbb.html)
12. [Business population estimates for the UK and regions 2025: statistical release](https://www.gov.uk/government/statistics/business-population-estimates-2025/business-population-estimates-for-the-uk-and-regions-2025-statistical-release)
13. [Enrich.chat agents.md](https://enrich.chat/agents.md)

## Try this prompt

Check an account list against your SAM

```text
Use www.enrich.chat to check my list of company websites against this SAM: US trucking companies with 200 to 999 employees. For each website, look up the company and return its industry, size band and state, then label it in or out of the SAM with one reason. For the first 40 that are in, find one VP of Operations. Mark any website with no match as not found, and report the total credits used.
```

## Related

- [Ideal customer profile template](https://enrich.chat/blog/ideal-customer-profile/)
- [Firmographic vs technographic data](https://enrich.chat/blog/firmographic-data/)
- [What is a B2B marketing database?](https://enrich.chat/blog/b2b-marketing-database/)