Market analysis · 13 min read · Updated 2 Aug 2026
TAM SAM SOM Analysis Template (Free Market Sizing)
A blank TAM SAM SOM analysis you can fill in today, plus the guidance for what belongs in each field. It is built bottom-up rather than from a published category total, because the number matters far less than whether anyone can reconstruct how you got to it.
Copy pastes straight into Google Sheets or Excel with the columns intact. Downloads are free with a work email.
The TAM SAM SOM analysis template
This is exactly what you get when you copy or download. Blank fields are yours to fill in; each table ships with one example row to show the pattern, which you delete.
Sizing scope
Fill this in first. Say who will read it, because a sizing for an investor and one for a roadmap decision get built to different standards.
- What we are sizingA product, a segment or a new market. Be specific
- AudienceInvestors, the board, or an internal build-or-not decision
- The decision this informse.g. whether the segment is large enough to justify a dedicated team
- Basis and periodAnnual recurring revenue, and the year it refers to
- Owner and dateOne named owner, with the date
- Primary methodBottom-up or top-down. Bottom-up unless you can say why not
1. What each layer means here
First row is an example, delete it. Complete all three rows before you calculate anything, and fill the exclusions column properly.
| Layer | Our definition for this analysis | What it deliberately excludes |
|---|---|---|
| ExampleTAM | Every company worldwide with 10 or more sales seats that could use a sales CRM | Companies below 10 seats, where the job is done in a spreadsheet and stays there |
2. TAM, built bottom-up
First row is an example, delete it. Show every input and its source. A total nobody can reconstruct is a number that will be quoted for years after its inputs stopped being true.
| Input | Value | Source | Confidence |
|---|---|---|---|
| ExampleCompanies worldwide with 10 or more sales seats | Your counted figure | Company registry data and industry association counts | Medium, the count is the weakest input in the chain |
3. SAM: what we can actually serve
First row is an example, delete it. Only list limits a customer would run into today. Planned expansions belong in a separate labelled scenario.
| Constraint | Effect on the number | Why it applies today | Would removing it be a project? |
|---|---|---|---|
| ExampleWe sell only in Europe | Reduces TAM to the European portion | No entity, no local support and no localised billing outside Europe | Yes, roughly a year and a funded programme |
4. SOM: what we can realistically win
First row is an example, delete it. Every input here should be a number you measured internally, and the period must be stated.
| Input | Value | Basis | Confidence |
|---|---|---|---|
| ExampleDeals our current sales capacity can close in the period | Reps multiplied by observed quota attainment | Our own closed-won data from the last four quarters | High, this is measured internally |
5. Assumption register
First row is an example, delete it. Every number above rests on assumptions. List them here so a reader can attack the weakest one rather than the conclusion.
| Assumption | Value used | Source | How sure are we? |
|---|---|---|---|
| ExampleAverage contract value in the target segment | Our observed figure | Closed-won deals in the segment, last 4 quarters | High for our segment, unknown outside it |
6. Sensitivity check
First row is an example, delete it. Vary the two or three inputs that move the answer most. If a plausible change flips the decision, the sizing is not yet usable.
| Assumption varied | Low case | Base case | High case | Effect on the decision |
|---|---|---|---|---|
| ExampleAverage contract value | 20% below observed | Observed figure | 20% above observed | Decision holds across the range, so it is not the binding assumption |
7. Cross-checks against reality
First row is an example, delete it. Test the numbers against something observable. Most sizing errors are order-of-magnitude and a single cross-check catches them.
| Cross-check | What it implies | Does it reconcile? |
|---|---|---|
| ExampleOur SOM against the largest competitor's estimated revenue in the same segment | If our SOM exceeds their actual revenue, an assumption is wrong | Yes, our SOM is a fraction of their estimated figure |
8. Decisions, owners and dates
First row is an example, delete it. Three to five rows. The most useful outcome is often that the market is smaller than assumed.
| Finding | What we will do | Owner | By when | How we will know it worked |
|---|---|---|---|---|
| ExampleSAM is far smaller than TAM because of one removable constraint | Cost the constraint removal as a project and decide separately | Elena V., CEO | 30 Nov | A funded decision exists rather than a recurring discussion |
How to fill in your TAM SAM SOM analysis
How to scope a TAM SAM SOM analysis
Say who will read it, because a sizing prepared for investors and one prepared for an internal build decision are held to different standards and answer different questions. Investors want to know whether the ceiling is high enough to matter; an internal decision needs to know whether the reachable portion justifies a team next year. The same three numbers serve both badly if the audience is not decided first.
What we are sizing
A specific product, segment or new market rather than the company. Sizing everything at once produces a headline figure and no decision, which is the most common outcome of this exercise.
Audience
Investors, the board, or an internal decision. Internal sizings should be conservative and inspectable, since somebody will staff a team against them. Investor sizings are frequently more ambitious, and it is worth being conscious of that rather than drifting into it.
The decision this informs
Name it: whether a segment justifies a dedicated team, whether a market entry is worth a year of engineering, whether to raise. A sizing with no decision behind it is a slide, and it will be argued about on aesthetics.
Basis and period
Annual recurring revenue for a stated year, usually. Mixing one-off and recurring revenue, or blending years, is a common and invisible error that makes the layers incomparable.
Primary method
Bottom-up unless you can articulate why not. Bottom-up numbers are smaller and defensible; top-down numbers are larger and collapse under the first serious question about how the category was defined.
How to define TAM, SAM and SOM for your business
Define the three layers for your business before calculating anything, because the generic definitions do almost no work. TAM is the total revenue if you had every possible customer. SAM is the portion your business model, geography and product can actually serve today. SOM is what you could realistically win in a defined period. Those sentences are agreed by everyone and settle nothing: all the difficulty is in where you draw each boundary and why.
TAM
Every organisation that could plausibly buy this category of product, worldwide, ignoring your own constraints. The exclusions matter more than the inclusions: companies too small to ever buy, or whose need is genuinely met by something free, are not addressable however much you would like them to be.
SAM
Narrowed by the constraints that genuinely apply today: geography where you can sell and support, segments your product actually fits, buyers your model can reach. Constraints you intend to remove next year do not belong here, they belong in a separate scenario clearly labelled as one.
SOM
What you can realistically win given capacity, competition and time. This is the number that should drive internal decisions and the one most often skipped in favour of the impressive one at the top.
What it deliberately excludes
Write the exclusions per layer. This column is what makes the analysis challengeable, and being challenged on a boundary is far better than being quietly disbelieved on the total.
Is TAM SAM SOM a framework?
Not in the sense that Porter's five forces is. It has no single originator and no founding text: it emerged as practitioner vocabulary and was popularised through startup literature, notably Steve Blank and Bob Dorf's The Startup Owner's Manual. That matters practically, because there is no authority to appeal to on where the boundaries go. You have to define them and defend them yourself.
How to calculate TAM bottom-up
Show every input and its source, because a total nobody can reconstruct is a number that will be quoted for years after its inputs stopped being true. Bottom-up means counting the addressable population and multiplying by a realistic revenue per customer, with every step visible. It produces a smaller number than a published category total and a far more defensible one, and the difference between those two properties is what makes it worth the extra work.
Count the population
Companies matching your addressable definition, from registries, industry associations, or platform data where a proxy exists. This is almost always the weakest input in the chain, so it deserves the most scrutiny and the most honest confidence rating.
Multiply by realistic revenue per customer
Use your own observed contract value rather than an aspirational price, and use the median rather than the mean if a few large deals distort it. Sizing built on list price when nobody pays list price is a common way to inflate a total by a third without noticing.
Show the arithmetic
Population times value, with any intermediate steps written out. A reader who can follow the calculation can improve it; one who cannot will either accept it uncritically or dismiss it, and both are worse.
Source every input
Named sources with dates. Where an input is your own estimate, say so rather than leaving it looking like external data, which is the most common way a sizing acquires false authority.
Treat published category totals with real caution
Market research reports frequently define categories differently from how buyers behave, rarely disclose methodology, and are often the origin of a figure everyone repeats without checking. Use one as a top-down cross-check if you like, never as the primary basis.
How to narrow TAM to SAM
One row per constraint that genuinely narrows the market today, not per constraint you plan to remove later. This is the section that determines whether the analysis is honest, because the temptation is to treat every limitation as temporary and therefore ignorable. A constraint you have a plan to remove is still a constraint until it is removed, and a SAM that assumes away geography, product gaps and business model limits is just TAM with extra steps.
Constraint
The real ones: geography where you can sell, bill and support; segments the product genuinely fits; buyer types your motion can reach; regulatory or compliance limits; languages you operate in. Each should be something a customer would run into, not something you would prefer to be true.
Effect on the number
Roughly how much it removes. Seeing that one constraint accounts for most of the gap between TAM and SAM is frequently the most actionable finding the exercise produces, because it converts a sizing question into a specific investment question.
Why it applies today
One line grounded in operational reality: no entity, no local support, no localised billing. Constraints stated without a reason invite endless argument about whether they are real.
Would removing it be a project?
The column that turns this table into a plan. A constraint removable in a quarter and one requiring a year and a funded programme are different animals, and the second deserves its own decision rather than an assumption inside a sizing.
Keep aspirational scenarios separate
If you want to show SAM after a planned expansion, present it as a clearly labelled second scenario alongside the current one. Blending the two produces a number that describes no year in particular.
How to calculate SOM realistically
Build this from capacity rather than from a share percentage, because a chosen percentage of SAM is a wish with a decimal point. The familiar move of asserting that you will capture some tidy fraction of the serviceable market is the least defensible step in a typical sizing, and it is usually the number a decision actually depends on. Capacity-based SOM is smaller, harder to argue with, and considerably more useful for planning.
Start from capacity
Sales headcount multiplied by observed quota attainment, or marketing-sourced pipeline multiplied by conversion rates you have measured. These are internal, measured and defensible, which makes them the right foundation.
Apply competitive reality
You will not win every deal you reach. Use your actual win rate against the competitive set in that segment, which you have from win/loss data, rather than an assumed one. Competitors' presence is the constraint most often omitted from SOM entirely.
Bound it by time
SOM is meaningless without a period. Obtainable in one year and obtainable in five are different numbers and the difference is often an order of magnitude, so state the horizon prominently.
Sanity-check against your own growth
If your SOM implies growing several times faster than you ever have, without a stated reason for the discontinuity, the number is aspiration rather than analysis. This single check catches most unusable SOMs.
Do not derive SOM as a percentage of SAM
Percentages have no mechanism behind them, so they cannot be defended or improved. If you must express SOM as a share for communication, compute it bottom-up first and derive the percentage afterwards, never the other way round.
How to keep an assumption register
Every number above rests on assumptions, and listing them lets a reader attack the weakest one rather than the conclusion. This is counter-intuitive if you are presenting a sizing you want believed, and it is what makes it believable. A total presented without its assumptions invites wholesale scepticism, because the reader has no way to distinguish the parts that are measured from the parts that are guessed.
Assumption
Written as a claim: "average contract value in the target segment is our observed figure". Everything that could be wrong belongs here, including the ones that feel obvious, since obvious assumptions are the ones nobody checks.
Value used
The specific number that went into the calculation, so a reader can substitute their own and see what happens. This is what makes the analysis a model rather than a conclusion.
Source
Internal data, published source, or estimate. Mark estimates clearly. A register where every row looks equally sourced defeats the purpose of having one.
How sure are we?
Per assumption, and specific about scope: "high for our segment, unknown outside it". Assumptions that hold in the segment you know and are extrapolated to one you do not are the most common failure in market sizing.
Rank by impact, not by number of rows
Two or three assumptions usually drive most of the answer. Identifying which ones those are is the whole purpose of the register, and it tells you where the next hour of research should go.
How to run a sensitivity check
Vary the two or three inputs that move the answer most, and if a plausible change flips the decision, the sizing is not yet usable. This is the step that separates a model from a number, and it is almost always skipped. It takes about twenty minutes in a spreadsheet and it routinely reveals that the entire analysis rests on one input nobody has verified, which is a considerably more useful finding than the total.
Assumption varied
Start with the highest-impact ones from the register: usually population count, average contract value and win rate. Varying inputs that barely move the answer is busywork that produces reassuring tables.
Low, base and high
Plausible ranges rather than dramatic ones. Plus or minus 20% on a measured input and a wider band on an estimated one is a reasonable default, and the ranges should be defensible in their own right.
Effect on the decision
The column that matters, and it is about the decision rather than the number. A sizing where the decision holds across the whole range is robust and you can stop. One where it flips has identified exactly what to go and verify.
Report the range, not just the base case
Presenting a single figure implies a precision the inputs do not support. A stated range with the base case marked is more honest and, in practice, taken more seriously by anyone who has built one of these before.
If everything is uncertain, the answer is a decision rule
When several assumptions swing the outcome, the useful output is not a number but a condition: proceed if the population count exceeds a threshold, otherwise do not. That is a defensible thing to bring to a decision meeting.
How to cross-check a market sizing
Test the numbers against something observable, because most sizing errors are order-of-magnitude and a single cross-check catches them. The classic failure is a SOM that quietly exceeds the actual revenue of the market leader, or a TAM implying that every company on earth buys three of these. Neither survives ten seconds of comparison against a known quantity, and neither is rare in documents that were never checked this way.
Against competitor revenue
If your SOM approaches or exceeds the estimated revenue of the largest player in that segment, an assumption is wrong. This is the fastest and most reliable cross-check available, and it needs only a rough competitor estimate.
Against your own history
Compare implied growth against what you have achieved. A SOM requiring five times your best year, with no stated reason for the discontinuity, is a forecast rather than an analysis.
Against the top-down figure
Where a published category total exists, compare it with your bottom-up TAM and state the gap. They will disagree; the size and direction of the disagreement is informative, and a bottom-up figure far larger than a published total usually means the population count is wrong.
Against unit economics
Multiply the implied customer count by what it costs to acquire and serve one. Sizings that are arithmetically fine but economically impossible are common, and this check is the one that surfaces them.
Record checks that failed
A cross-check that did not reconcile, and what you changed as a result, is the strongest evidence a reader has that the analysis was done carefully. Removing the failed check and keeping the corrected number throws that evidence away.
How to fill in the decisions section of your market sizing
Three to five rows, and the most useful outcome is often that the market is smaller than assumed. That is an uncomfortable finding to deliver and a valuable one to have before staffing a team against the optimistic version. Sizings that always conclude the opportunity is large enough are sizings whose conclusion was decided first, and everyone reading them senses it even when they cannot point at the specific step.
Finding
Specific and drawn from the tables: "SAM is far smaller than TAM because of one removable constraint". Findings about the structure of the number are more actionable than the number itself.
What we will do
Frequently one of three: proceed, cost a constraint removal as a separate funded decision, or verify the binding assumption before deciding. The third is the right answer more often than teams expect and is the cheapest of the three.
Owner
Named. Sizing outputs are unusually prone to being interesting to everyone and owned by nobody, since they typically imply work across product, sales and finance.
By when
Tied to the decision the sizing was built for. A sizing delivered after the investment decision was taken is an artifact of the process rather than an input to it.
How we will know it worked
Usually the existence of a funded decision rather than a market outcome, since market size does not change in response to anything you do. The test is whether the analysis ended an argument rather than continuing it.
Sourcing and upkeep: keeping a market sizing defensible
These rules apply to every section above. This artifact has the weakest reputation of anything in this cluster, and largely deserves it: market sizing is routinely performed backwards, with a desired conclusion chosen first and inputs assembled until they support it. The habits below are what separate a sizing that survives scrutiny from one that produces a large number and no confidence.
Bottom-up, always, with the arithmetic visible
Count the population and multiply by observed value. Smaller and defensible beats larger and unreconstructable, and a reader who can follow the calculation can improve it.
Never derive SOM as a percentage of SAM
A chosen fraction has no mechanism behind it and cannot be defended. Build it from sales capacity and measured win rates, then derive the percentage afterwards if you need it for communication.
Keep aspirational constraints out of SAM
A limitation you plan to remove is still a limitation. Present the post-expansion view as a clearly labelled second scenario rather than blending it in.
Publish the assumption register
It lets a reader attack the weakest input rather than the conclusion, which is both more useful and, counter-intuitively, what makes the analysis credible.
Run the sensitivity check
Twenty minutes, and it routinely reveals that the whole answer rests on one unverified input. If a plausible variation flips the decision, go and verify that input before presenting anything.
Cross-check against a known quantity
Competitor revenue, your own growth history, unit economics. Most sizing errors are order-of-magnitude and one comparison catches them. Record the checks that failed and what you changed.
Re-size when the business changes, not annually
Market sizes move slowly; your constraints do not. The trigger for redoing this is entering a geography, shipping something that widens product fit, or changing the motion, rather than the calendar.
A TAM SAM SOM analysis example
You are at Pipedrive, deciding whether the mid-market segment justifies a dedicated team next year. The board wants a market size, and the useful version is bottom-up with the working shown. This is that analysis, filled in.
Published pricing and packaging verified 2 August 2026, from the companies’ own pages rather than third-party round-ups, which frequently conflate annual and monthly prices. Pricing changes without notice, so re-check before quoting any of it.
Sections marked illustrative are invented for this example. Win rates, deal counts, discounting behaviour, customer quotes, owners and internal dates are not published by HubSpot, Pipedrive or anyone else, so those rows are a plausible fictional scenario rather than reported fact, and should not be read as claims about how either company performs or negotiates. Everything else comes from the two pricing pages linked below, read on the date shown.
Sizing scopeIllustrative
| Field | Example entry |
|---|---|
| What we are sizing | Sales CRM for teams of 25 to 100 seats, as a standalone segment |
| Audience | Internal, for a build-or-not decision on a dedicated segment team |
| The decision this informs | Whether the segment justifies a dedicated team of six next year |
| Basis and period | Annual recurring revenue, calendar 2027 |
| Owner and date | Tom A., Competitive Intelligence, with Finance, 2 Aug 2026 |
| Primary method | Bottom-up. A published total is used only as a cross-check |
1. What each layer means hereIllustrative
| Layer | Our definition for this analysis | What it deliberately excludes |
|---|---|---|
| TAM | Every company worldwide with 25 to 100 sales seats that could use a sales CRM | Companies below 25 seats and above 100, both of which buy differently and are separate questions |
| SAM | The European portion of that, in languages we support, excluding regulated sectors we cannot serve | Every geography where we have no entity, no local billing and no support coverage |
| SOM | What our current and planned sales capacity can realistically close in calendar 2027 | Anything requiring headcount beyond the plan, or a win rate above what we have measured |
2. TAM, built bottom-upIllustrative
| Input | Value | Source | Confidence |
|---|---|---|---|
| Companies worldwide with 25 to 100 sales seats | Our counted figure | Company registry data cross-referenced with industry association counts | Medium. This is the weakest input in the whole chain |
| Share of those that use or would use a sales CRM | A stated percentage, not 100% | Our own discovery data on what prospects use today | Low to medium. Derived from deals we entered, so it is biased |
| Average annual contract value at this seat range | Our observed median, not list price | Closed-won deals in the segment, last 4 quarters | High for Europe, unknown elsewhere |
| TAM | Population times CRM adoption times contract value | The three rows above, multiplied | Medium, and no better than its weakest input |
3. SAM: what we can actually serveIllustrative
| Constraint | Effect on the number | Why it applies today | Would removing it be a project? |
|---|---|---|---|
| We sell only in Europe | Reduces TAM to roughly the European portion, the largest single cut | No entity, no localised billing and no support coverage elsewhere | Yes. Roughly a year and a funded programme per major region |
| We support six languages | Removes a further slice of the European population | Product and support are not localised beyond those six | Partly. Two more languages would be a quarter of work |
| We cannot serve regulated sectors requiring specific certifications | Removes a modest but real slice | We do not hold the certifications and the audit cycle is long | Yes, and it is a multi-quarter compliance programme |
| Our product genuinely fits teams without dedicated ops headcount | Removes larger, more complex buyers inside the seat range | Deals needing heavy customisation churn or never close | No. This is a deliberate product position, not a gap |
4. SOM: what we can realistically winIllustrative
| Input | Value | Basis | Confidence |
|---|---|---|---|
| Reps carrying a mid-market quota in 2027 | Current headcount plus the approved plan | The approved hiring plan | High, it is a decision already taken |
| Observed quota attainment | Our measured average, not target | Last four quarters of attainment data | High, measured internally |
| Win rate against the competitive set in this segment | Our measured rate, currently around a third | Win/loss data on tagged competitive deals | Medium. Competitor tagging covers about 71% of closed deals |
| SOM for 2027 | Reps times attainment, adjusted for win rate | The three rows above | Medium, and deliberately conservative |
5. Assumption registerIllustrative
| Assumption | Value used | Source | How sure are we? |
|---|---|---|---|
| Average contract value at 25 to 100 seats holds in 2027 | Our observed median | Closed-won, last 4 quarters | High for our segment, and it assumes no pricing change |
| The company population count is roughly right | Our counted figure | Registry data, cross-referenced | Medium. It is the input we have verified least |
| Win rate stays near a third | Our measured rate | Win/loss on tagged deals | Medium. It fell 7 points last quarter against one competitor |
| Quota attainment does not degrade as we hire | Current average | Attainment data | Low. New reps ramp, and the plan adds several at once |
| No major competitor packaging change during 2027 | Assumed stable | None. This is an assumption, not a finding | Low, and our own war game suggested a match within a quarter |
6. Sensitivity checkIllustrative
| Assumption varied | Low case | Base case | High case | Effect on the decision |
|---|---|---|---|---|
| Average contract value | 20% below observed | Observed median | 20% above observed | Decision holds across the range. Not the binding assumption |
| Win rate | Down 10 points | Around a third | Up 5 points | Decision flips at the low end. This is the binding assumption |
| Quota attainment as headcount grows | Degrades 20% during ramp | Holds flat | Improves slightly | Decision holds but the payback year moves by two quarters |
| Company population count | 30% below our figure | Our figure | 30% above | Decision holds, because SOM is capacity-bound rather than market-bound |
7. Cross-checks against realityIllustrative
| Cross-check | What it implies | Does it reconcile? |
|---|---|---|
| Our SOM against the largest competitor's estimated revenue in this segment | If SOM approached their revenue, an assumption would be wrong | Yes. Our SOM is a fraction of their estimated figure |
| Implied growth against our own history | SOM implies growth in line with our best two years, not beyond | Yes, and this is why the capacity-based method was used |
| Bottom-up TAM against a published European CRM total | They disagree by roughly 30%, with ours lower | Partly. We report the gap rather than reconciling it |
| Unit economics on the implied customer count | Acquisition and service cost at that volume against contract value | Failed initially. The first SOM implied a cost to serve we could not fund, and the capacity input was corrected |
8. Decisions, owners and datesIllustrative
| Finding | What we will do | Owner | By when | How we will know it worked |
|---|---|---|---|---|
| The decision is bound by win rate, not by market size | Stop refining the population count and fix the read-only seat gap instead | Elena V., CEO | 31 Oct 2026 | The pricing decision is taken before the team is staffed |
| Geography is the largest single cut between TAM and SAM | Cost one region entry as a separate funded decision rather than assuming it | Elena V., CEO | 30 Nov 2026 | A funded decision exists rather than a recurring discussion |
| Ramp degradation is the weakest assumption in the register | Model the hiring plan with a ramp curve before committing to six | Finance, with Dana K., Sales | 30 Nov 2026 | The plan uses a ramp curve rather than flat attainment |
| One cross-check failed and changed the answer | Keep the failed check in the document rather than removing it | Tom A., Competitive Intelligence | 2 Aug 2026 | The board sees what was corrected and why |
How to roll out your TAM SAM SOM analysis
- 1Copy or download the blank analysis. Use Copy to paste it straight into Google Sheets or Excel with the columns intact, or download the CSV, Notion or PDF version.
- 2Decide the audience and the decision first. An investor sizing and an internal build decision are held to different standards, and the same three numbers serve both badly if you do not choose.
- 3Delete the example rows. Each table ships with one example row so the pattern is obvious. Remove it before you circulate the analysis.
- 4Define the three layers for your business, with exclusions. The generic definitions settle nothing. All the difficulty is in where you draw each boundary and why, so write the exclusions down.
- 5Build TAM bottom-up with the arithmetic visible. Count the population, multiply by observed contract value rather than list price, and source every input including your own estimates.
- 6Narrow to SAM using constraints that apply today. A limitation you plan to remove is still a limitation. Keep post-expansion views as a separate labelled scenario.
- 7Build SOM from capacity and measured win rates. Never as a chosen percentage of SAM. Bound it by a stated period, and sanity-check the implied growth against your own history.
- 8Run a sensitivity check, then cross-check against reality. Vary the two or three inputs that move the answer most. Compare SOM against the leader's estimated revenue, and record checks that failed.
TAM SAM SOM analysis FAQ
What is TAM SAM SOM?
Three nested estimates of market size. TAM, the total addressable market, is the revenue available if you had every possible customer. SAM, the serviceable addressable market, is the portion your business model, geography and product can actually serve today. SOM, the serviceable obtainable market, is what you could realistically win in a stated period. The definitions are uncontroversial and settle nothing: all the difficulty, and all the disagreement, is in where you draw each boundary and whether you show your working.
What is the difference between TAM, SAM and SOM?
Each narrows the one above by a different kind of constraint. TAM to SAM is narrowed by what you can serve: geography, language, product fit, business model, regulatory reach. SAM to SOM is narrowed by what you can win: sales capacity, win rate against the competitive set, and the time period. The most common error is blurring the two, typically by treating a constraint you plan to remove next year as though it were already gone, which produces a SAM that is really TAM with extra steps.
How do you create a TAM SAM SOM?
Bottom-up. Count the companies matching your addressable definition, multiply by a realistic contract value from your own closed-won data rather than list price, and show every step. Narrow to SAM using constraints that genuinely apply today, one row each with its effect. Build SOM from sales capacity and measured win rates rather than as a percentage of SAM. Then list your assumptions, vary the two or three that move the answer most, and cross-check the result against a known quantity such as the market leader's estimated revenue.
Is TAM SAM SOM a framework?
Not in the way Porter's five forces or the Competitive Profile Matrix are. It has no single originator and no founding text: it emerged as practitioner vocabulary in venture and startup circles and was popularised through startup literature, notably Steve Blank and Bob Dorf's The Startup Owner's Manual. It is better understood as a shared way of labelling three levels of market estimate than as a method with rules. That has a practical consequence: there is no authority to appeal to on where the boundaries belong, so you have to define them explicitly and defend them yourself.
How do you calculate SOM for a startup?
From capacity, never as a percentage of SAM. Take the number of reps carrying quota in the period, multiply by observed quota attainment rather than target, and adjust by your measured win rate against the competitive set in that segment. For a self-serve motion, substitute marketing-sourced pipeline and measured conversion rates. Then sanity-check the implied growth against your own history: if the result requires several times your best year with no stated reason for the discontinuity, it is a forecast rather than an analysis.
What is a good SOM percentage?
The question is the wrong way round, and answering it is how most sizings go wrong. SOM should be built bottom-up from capacity and measured win rates, and any percentage of SAM derived afterwards for communication. A chosen percentage has no mechanism behind it, so it cannot be defended when questioned or improved when you learn something. If you are asked what percentage is credible, the honest answer is that it depends entirely on your capacity and competitive position, and that a number arrived at by choosing a percentage first should be distrusted regardless of its size.
What is a good TAM size?
There is no threshold that makes a TAM good, and the question usually stands in for whether investors will find a market interesting, which depends on the fund, the stage and the thesis rather than on an absolute figure. More useful: a TAM you can reconstruct is worth far more than a large one you cannot. A defensible bottom-up figure that is smaller than a rival's inflated top-down number will survive diligence, and the inflated one will not. Optimise for showing your working rather than for the size of the total.
How do you do market sizing?
Two methods exist and they are not equal. Bottom-up counts the addressable population and multiplies by realistic revenue per customer, with every input visible and sourced. Top-down starts from a published category total and narrows it by segment and geography. Bottom-up is smaller, slower and defensible; top-down is faster, larger and collapses under the first serious question about how the category was defined. Use bottom-up as your primary method and top-down only as a cross-check, then state the gap between them rather than quietly picking one.
What is the difference between PAM and TAM?
PAM, usually expanded as potential available market, appears occasionally as a layer above TAM covering the market that could exist if constraints such as regulation, technology or adjacent categories changed. It is not part of the standard vocabulary: most treatments use only the three layers, and adding a fourth tends to widen the top of the funnel without adding decision value. If someone in your organisation uses PAM, ask what it excludes that TAM includes, because the answer varies considerably between people using the same word.
Is TAM SAM SOM per year?
Almost always annual, and it should be stated explicitly because the layers become meaningless if mixed. TAM and SAM are normally expressed as annual revenue if fully captured. SOM is bounded by a specific period, and this is where the ambiguity usually sits: obtainable in one year and obtainable in five are different numbers, often by an order of magnitude. If a sizing does not state the SOM horizon, that is the first question to ask, and the answer frequently reveals that nobody decided.
Who uses TAM SAM SOM?
Founders raising capital, investors assessing whether a market is large enough to return a fund, and internal teams deciding whether a segment justifies investment. Those are genuinely different uses and they warrant different standards. An investor sizing is arguing that a ceiling is high enough to matter. An internal sizing will have a team staffed against it, so it should be conservative, inspectable and honest about which assumption binds. Using the same document for both is common and usually serves the internal decision badly.
What is the difference between TAM SAM SOM and market share analysis?
Direction. Market sizing looks forward at an opportunity and estimates what portion you might capture. Market share analysis looks backward at a market that exists and measures what proportion each player currently holds. They share the hardest problem, defining the market honestly, and they are frequently confused in board material. A useful discipline: use share when reporting performance and sizing when justifying investment, and never present one as evidence for the other, since they rest on different assumptions.
What does SOM mean in audit?
A different thing entirely, and worth separating since the acronym collides. In auditing, SOM commonly refers to a summary of misstatements, the schedule an auditor keeps of identified errors and their aggregate effect on the financial statements. It has no connection to serviceable obtainable market. Similarly, the SAM model in a sales context usually refers to something unrelated to serviceable addressable market. Acronyms in this area overlap heavily across disciplines, so it is worth confirming which sense is meant before answering.
How reliable are published market size reports?
Treat them with real caution as a primary basis. Category definitions in published reports rarely match how buyers actually shop, methodologies are frequently undisclosed, and a surprising number of widely-cited figures trace back to a single estimate that has been repeated until it acquired authority. They are useful as one top-down cross-check against a bottom-up number you built yourself. They are not a substitute for it, and a sizing whose only support is a purchased report will not survive diligence by anyone who asks how the category was defined.
What should a sensitivity check cover?
The two or three assumptions that move the answer most, which are usually the population count, average contract value and win rate. Vary each across a plausible range, typically plus or minus 20% on a measured input and wider on an estimated one, and record the effect on the decision rather than on the number. If the decision holds across the whole range, the sizing is robust and you can stop refining it. If a plausible variation flips it, you have found exactly what to verify before presenting anything.
How do you cross-check a market sizing?
Against something observable. Compare your SOM with the largest competitor's estimated revenue in the same segment: if yours approaches theirs, an assumption is wrong. Compare implied growth against your own best years. Compare your bottom-up TAM with a published total and state the gap. And check the unit economics, since a sizing can be arithmetically sound and economically impossible. Most sizing errors are order-of-magnitude, so a single comparison against a known quantity catches them, and it takes minutes.
How often should you redo a market sizing?
When the business changes rather than on a calendar. Market sizes move slowly; your constraints do not. The triggers worth acting on are entering a new geography, shipping something that widens product fit, changing the sales motion, or a pricing change that alters average contract value. Redoing it annually out of habit produces a document nobody reads, while redoing it when a constraint disappears produces a genuinely different SAM and frequently a different decision.
What is an example of a TAM SAM SOM analysis?
Sizing the 25 to 100 seat sales CRM segment for an internal decision about whether to staff a dedicated team of six. TAM built bottom-up: companies worldwide in that seat range, multiplied by the share that use or would use a CRM, multiplied by observed median contract value rather than list price, with each input sourced and the population count flagged as the weakest link. SAM narrowed by four constraints that apply today, the largest being that we sell only in Europe, each recorded with whether removing it would be a funded project. SOM built from reps carrying quota, observed attainment and measured win rate, deliberately conservative. The sensitivity check produced the finding that mattered: the decision holds across plausible ranges for contract value and population count, and flips only on win rate, which means the analysis was never really about market size. One cross-check failed, since the first SOM implied a cost to serve the company could not fund, and the corrected version stayed in the document rather than being quietly removed. Decision: stop refining the population count and fix the packaging gap that is costing win rate.
What are the most common mistakes in TAM SAM SOM?
Six recur. Starting top-down from a purchased report and treating it as fact. Deriving SOM as a chosen percentage of SAM, which has no mechanism and cannot be defended. Treating constraints you plan to remove as already removed, which turns SAM into TAM with extra steps. Using list price rather than observed contract value, which inflates every layer at once. Presenting a single figure with no range, implying precision the inputs cannot support. And never cross-checking against a known quantity, which is how a SOM larger than the market leader's actual revenue reaches a board deck.
TAM SAM SOM analysis grounded in competitor data
Flares tracks competitor pricing, packaging and segments, so your market sizing rests on observed evidence.
Discover Flares14-day free trial · 30-second setup