ProjekinAja. Founder's Playbook Stage 1 · Signal or Noise Lesson 1.7
● Stage 1 · Lesson 1.7

Market Sizing Without Lying to Yourself

TAM/SAM/SOM as commonly practiced is venture theater — here is how to build a credible bottom-up market estimate that survives a skeptical room.

⏱ ~20 min read 📐 Frameworks & numbers 🧮 Interactive calculator

🎯 By the end you'll understand

Why top-down market sizing destroys credibility with sophisticated investors, how to construct a defensible bottom-up estimate starting from reachable ICPs, which assumption drives your final number most (and why that is actually the highest-risk assumption in your business), and how to avoid both lethal sizing errors — the market too small to fund, and the market too large to be taken seriously.

KEY INSIGHTTAM slides do not size markets — they reveal how the founder thinks

A $50B TAM tells an investor nothing useful about your business. A bottom-up estimate that derives $40M from 1,200 reachable plants × 12% conversion × $28K ACV tells them you understand your market, your customer, and your own assumptions — and gives them three specific things to argue with.

Here is the slide every first-time founder shows: "Global industrial IoT market: $263 billion by 2027, growing at 23% CAGR. We are targeting a 1% share, which gives us a $2.6B opportunity." An experienced investor has seen this slide — or something functionally identical — several hundred times. The "1% of a big number" move is a tell. It signals that the founder does not actually know how many customers exist, how many are reachable, what they would pay, or how many they can realistically close. The big TAM was borrowed from a market research report; the 1% was invented to produce an impressive-sounding number. Neither figure is specific to the startup's actual opportunity.

🔑 What market sizing is actually for

Market sizing serves three purposes, none of which require a giant top-down number. First, it tells you whether the market is large enough to build a venture-scale (or lifestyle-scale, or acqui-hire-scale) business. Second, it helps you discover which assumption about your market is most fragile — the one that, if wrong, collapses the opportunity. Third, it gives investors a structured argument to engage with. A bottom-up model invites productive disagreement ("I think your conversion rate is high"); a top-down number invites dismissal ("that's not a real analysis"). Build the one that starts a real conversation.

CORE CONCEPTThe top-down TAM trap — and why it backfires

Top-down sizing starts from a macro number — typically sourced from an industry research report — and narrows down via assumed market share. The logic sounds plausible at first: the global quality-control-systems market is $15B; automotive accounts for roughly 22% of industrial manufacturing; predictive maintenance is about 30% of that; therefore our addressable market is ~$1B. We only need 2% to be a great business.

The problem is structural, not arithmetic. Every step in that chain is borrowed from a report that was built for different purposes, aggregated at the wrong level, and applied to a customer segment you have not actually defined. The real question is not "what fraction of the global market report could we theoretically address?" but "how many actual buyers exist, where are they, what do they pay, and how many can we reach with our actual go-to-market?"

Top-Down (Theater) 1% $263B global IoT report (generic, wrong level) × assumed sub-segment % (borrowed, not validated) invented share → impressive number Bottom-Up (Defensible) Reachable ICPs (countable plants) × Realistic conversion rate × Validated ACV = SAM you can actually defend
Top-down sizing borrows credibility from macro reports and applies an invented share percentage. Bottom-up builds from countable, reachable buyers and multiplies up — every input is something you can argue from first principles.

Top-down TAM has one legitimate use: establishing category magnitude. Saying "the industrial inspection software market is at least $8B globally" tells an investor the category is real and large enough to support multiple winners. That is useful framing. But when founders try to derive their actual revenue opportunity from top-down numbers, the analysis breaks down — because they never asked "how many of these buyers can I actually reach, convince, and close in the next 18 months?"

⚠ The size illusion that kills credibility

The "1% of a big market" framing is specifically dangerous because it makes the market large enough to look exciting while requiring zero work to justify. Any market can be made to look like a billion-dollar opportunity if you start from a sufficiently aggregate number. Investors who have seen this pattern know it means the founder has not done the hard work of counting buyers. The slide designed to impress instead signals that the founder does not know their customer.

HOW IT WORKSThe bottom-up construction method

Bottom-up sizing starts from the number of reachable ideal customers and builds up. The formula has three components and every component is something you should be able to defend from your discovery work:

Component What it is How to estimate it What "wrong" looks like
Reachable ICP count The number of companies (or facilities) that match your ICP profile and are reachable by your actual go-to-market motion Industry databases, NAICS codes, trade association member lists, LinkedIn company search, your own discovery pipeline Using total industry count (all manufacturers) instead of the specific tier, size, and segment that matches your actual ICP
Realistic conversion rate The fraction of reachable ICPs you can realistically convert to paying customers over a defined time horizon Comparable SaaS benchmarks for enterprise, adjusted for your sales motion; should feel uncomfortable, not optimistic Using 20–30% when comparable enterprise companies with established sales teams achieve 2–8% in year 1–3
ACV (Annual Contract Value) What each customer will pay annually, net of discounts and one-time implementation fees From your discovery interviews — what are they currently paying for the problem? What did they indicate willingness to pay? What is the value captured relative to your cost? Assuming enterprise ACV with no evidence of enterprise willingness to pay; or pricing to a cost-based floor without evidence of the customer's value calculation

The relationship between these three components gives you the three market estimates that actually matter for an early-stage company:

SAM

Serviceable Addressable Market — your real ceiling

SAM = Reachable ICP count × ACV. This is the maximum revenue you could generate if you converted every reachable ICP. It is your ceiling, not your target. A healthy SAM is at least 10× your 3-year revenue target — otherwise you will hit the ceiling before you scale, which limits both the business and the fundraising narrative. For a venture-backed business, SAM typically needs to be $500M+ to support a meaningful outcome; for a bootstrapped or PE-backed business, a $50M SAM can be excellent.

SOM

Serviceable Obtainable Market — your realistic 3-year target

SOM = Reachable ICP count × Conversion rate × ACV. This is what you can actually close in a realistic planning horizon. If your conversion rate is 5% and you have 2,000 reachable ICPs at $30K ACV, your SOM is $3M — which means your 3-year ARR target should be in that range. If your model needs $10M ARR to reach the next funding milestone but your SOM calculation shows $3M, you have a problem that no amount of hustle resolves; you need to either expand the ICP count, raise ACV, or revisit your conversion assumption.

TAM

Total Addressable Market — useful only for category calibration

TAM = Total market count × ACV. Use TAM to establish that the category is large enough for a venture-scale outcome, not to derive your target revenue. A bottom-up TAM uses your ACV applied to the total potential customer universe — not the entire industrial IoT market, but all facilities in your specific segment globally. This gives a number that is large enough to be interesting but small enough to be credible.

💡 The assumption that matters most

In almost every early-stage market-size model, ACV is the highest-leverage assumption. Doubling ACV doubles every number in your model. This is why pricing discipline (covered in Stage 3) is so important — founders who undercharge do not just leave money on the table, they also make their market look smaller than it is, which changes the fundraising narrative and the required growth rate to reach milestones. If your market model looks marginal, look at your ACV assumption first.

SEE ITMarket size calculator — top-down vs. bottom-up for Forge

⚙ Interactive · Forge Market Size Calculator

Build side-by-side top-down and bottom-up estimates for Forge. Set each assumption using the sliders. The widget derives SAM and SOM for both methods, shows the gap, highlights which assumption has the most leverage on the final number, and reveals whether each bottom-up assumption is defensible given what Forge's discovery work has found.

📊 Top-Down (Theater)

$18B
18%
12%
1.0%
Top-Down TAM (automotive CV inspection)
—
Your "addressable opportunity" (TAM × capture %)
—
—

🎯 Bottom-Up (Defensible)

1,200
4.0%
$32K
1.3×
Bottom-Up SAM (all reachable × ACV)
—
Bottom-Up SOM (conversion × ACV × expansion)
—
—
—

Bottom-up assumption defensibility — based on Forge's discovery work

Sensitivity analysis — which assumption drives the final number most?

Each bar shows how much the SOM changes with a ±25% move in that assumption, holding others fixed. The longest bar is the assumption most worth arguing about — and the most worth getting right.

📋 Market size verdict

Adjust the sliders to see the verdict.

Notice what the gap reveals. A typical founder's top-down slide shows a $300M–$600M "opportunity" from a 1% share of a large market report. The bottom-up analysis — which forces you to count actual plants, estimate a realistic conversion rate, and validate ACV — typically lands significantly lower, in the $30M–$80M SOM range for a North America-focused early-stage company. This is not a bad number; it is an honest one. And it is the number a sophisticated investor can engage with, stress-test, and potentially believe.

Also notice the sensitivity bars. Because SOM is purely multiplicative — every input enters the formula with equal weight — a 25% improvement in ACV has the same leverage as a 25% improvement in plant count, conversion rate, or expansion factor. None of these assumptions is uniquely dominant; they are equally load-bearing. The lesson: undercharging (a low ACV) is as damaging as missing your conversion target or overcounting your ICP — not because ACV dominates, but because every input multiplies the others. Founders who undercharge to accelerate adoption are not just leaving money on the table; they are making their market model look smaller, which requires higher growth rates to reach the same milestone, which increases execution risk across the entire company.

IN PRACTICEHow investors stress-test market sizing

When a Series A investor looks at your market size slide, they are running a specific set of checks — usually quickly and often without telling you. Understanding the checks helps you build a model that survives them:

1

The denominator check — where did this ICP count come from?

An investor will ask — or wonder — whether you counted the ICPs yourself or borrowed a number from a market report. "1,200 Tier-1 automotive supplier plants in North America" is a specific, checkable number. You can point to NAICS code 3363 (motor vehicle parts manufacturing), add NAICS 3362 (motor vehicle body/trailer manufacturing), cross-reference with EPA facility data or industry databases, and arrive at a defensible count. "The automotive sector represents $X billion" is not checkable in the same way. Do the denominator work; it pays for itself in credibility.

2

The conversion rate reality check

Enterprise SaaS with a direct sales motion typically sees 3–8% of pipeline convert to closed-won in early years, with significant variation by deal size and sales cycle length. Hardware + software bundles with complex integration (like an edge sensor requiring line installation) tend toward the lower end. A 15% conversion rate in year 1 for a hardware-augmented enterprise product with an 8–12 month sales cycle is a red flag. Investors apply their portfolio comparables here. Use 3–6% for a defensible assumption and explain what would have to be true to achieve the upper end.

3

The ACV anchoring question

Where did the ACV come from? "We plan to charge $32K" is not an answer. "We found in discovery that plants are currently spending $14K–$22K per line per year on inspection labor plus $8K–$15K on downstream rework attributable to late defect detection, totaling $22K–$37K. We are pricing Forge at $28K–$36K to capture roughly half the value we deliver" is an answer. The ACV should be anchored to something real — existing spend, value delivered, comparable product pricing, or willingness-to-pay signals from discovery. An investor who cannot see the anchoring will assume you invented it.

4

The "how does this number grow?" question

A static SOM is insufficient. Investors want to see that the market model has natural expansion mechanics — geographic expansion (North America → Europe → Asia), vertical expansion (automotive → aerospace → electronics manufacturing), product-line expansion (inspection → predictive maintenance → quality analytics), and customer expansion (one line per plant → whole facility). The expansion rate slider in the calculator above captures the intra-customer expansion; the stage-level expansion requires a separate narrative that builds on the initial SOM without double-counting.

FORGESizing Forge's actual market after Stage 1 discovery

At the end of Stage 1's 20 discovery interviews, Forge has real numbers to work from. The discovery work found that Tier-1 automotive suppliers lose $180K–$400K per unplanned downtime event, and that current inspection is manual, lagging, and a documented source of quality escapes. Here is how those findings translate into a defensible market model:

🏭 Forge's bottom-up market model — post-Stage 1

Reachable ICP (North America, Year 1–2 focus): Approximately 1,100–1,400 Tier-1 automotive supplier facilities in the U.S. and Canada with 500+ employees and active stamping, casting, or assembly lines — the segment where downtime cost is highest and manual inspection is most clearly inadequate. This estimate is built from NAICS 3363/3362 data and trade association membership rolls, cross-referenced against the plants Forge's interviews were drawn from.

ACV range: $28K–$40K per facility per year, anchored to the economic value of preventing 0.5 unplanned downtime events annually. At $180K average downtime cost, preventing half an event is worth $90K — Forge prices to capture roughly 30–40% of that value. Discovery interviews found willingness to consider pilots at $8K–$15K, consistent with a $28K–$40K annual figure once value is demonstrated.

Conversion rate (Year 1–3): 3–5%. This is conservative for enterprise hardware with an 8–12 month sales cycle. It implies 33–70 facilities closed in years 1–3, requiring a focused direct-sales motion with pre-warmed champions from the discovery pipeline.

SAM: 1,200 plants × $34K ACV = $40.8M. Reasonable for a venture-backed Series A story; not a category-defining number but defensible and honest.

SOM (3-year): 1,200 × 4% × $34K × 1.3× expansion = $2.1M–$2.7M ARR. This is consistent with a successful seed-to-Series-A story where Forge converts 48–60 facilities to paying customers at meaningful ACV before raising its A.

The Stage 1 go-signal is not a market-size question — it is a problem-and-signal question. But the market-size work that Forge does in Stage 1 (as part of completing this lesson's framework) serves a specific function: it establishes that the opportunity is large enough to pursue a venture path, while being honest enough not to collapse under the first investor stress-test. A $40M SAM with clear expansion mechanics is a defensible, fundable market story.

🔑 What the discovery interviews actually contributed to market sizing

The ICP profile from Lesson 1.3 gave Forge the denominator — the specific firmographic filter that defines "reachable." The pain quantification from Lessons 1.4–1.6 gave Forge the ACV anchor — the economic value that justifies the price. Without the discovery work, market sizing is entirely invented. With it, every number in the bottom-up model has a story. That is the difference between venture theater and a real market analysis.

~1,200 Tier-1/2 auto plants (NA) × $34K ACV anchored to value delivered = SAM $40.8M your ceiling if you close everyone × 4% conv. = SOM ~$2.1M 3-yr ARR target From: ICP definition (Lesson 1.3 + NAICS data) From: discovery interviews (Lessons 1.4–1.6)
Forge's bottom-up market model. Every input has a source in the discovery work — the count from ICP definition, the ACV from pain quantification, the conversion rate from enterprise benchmarks. This is a model that survives a skeptical room.

WEAKNESSWhere this bites — the two lethal sizing errors

✗ The market too small to fund — and the fix that makes it worse

When a bottom-up analysis reveals a $15M–$20M SAM, many founders' first instinct is to expand the ICP definition until the number is large enough. This is almost always the wrong move. Expanding the ICP to fix the market-size number is the cart pulling the horse: you are not discovering that your real market is larger, you are manufacturing a larger-looking number by making the ICP fuzzier. The right response to a small SAM is to ask whether the expansion is real — is there genuine evidence that adjacent segments have the same pain at the same severity? — or whether you are doing top-down reasoning in bottom-up clothing. A $20M SAM with tight product-market fit is often a better business than a $100M SAM with a sprawling ICP that no sales motion can serve.

✗ The market too large to be credible — and what it signals

The inverse error is just as damaging: claiming a $500M SAM when your ICP definition implies a much smaller reachable universe. This typically happens when founders use total industry count (all manufacturers globally) rather than the specific segment their ICP actually describes. An investor who asks "how many customers is that?" and gets back "well, there are 400,000 manufacturing facilities in North America" — and then asks "but how many meet your specific Tier-1 automotive supplier with 500+ employees criterion?" — and gets "we haven't actually counted" — has learned that the market model is theater. Claiming a market size you cannot defend numerically is worse than claiming a smaller, defensible one.

✗ Using market size to validate what should be validated by discovery

Market size is not evidence of problem severity. A $40B global market tells you the category is large; it tells you nothing about whether your specific ICP has a specific, urgent, underfunded problem. Founders sometimes use a large TAM as a substitute for discovery evidence — "the market clearly needs this, look how big it is." Investors see through this immediately. Market size and problem validation are separate questions. The market size work in this lesson assumes you have already completed the discovery work in Lessons 1.3–1.6 and are now sizing a market you have actually validated. Doing the sizing first, then the discovery, inverts the logical order and typically produces a motivated analysis rather than an honest one.

KEY INSIGHTHow this connects

  • 1.3 · Ideal Customer Profile — The Behavioral Portrait — the ICP definition is the denominator of your bottom-up model. Without a precise ICP, you cannot count reachable buyers; without a count, your SAM is invented. The behavioral specificity of a good ICP (industry, size, role, trigger, behavior) is what makes the reachable ICP count checkable rather than estimated.
  • 1.6 · Reading Signal — Real Pain vs. Polite Interest — pain quantification from structured discovery is what anchors the ACV. If you know that unplanned downtime costs Tier-1 plants $180K–$400K per event, and that Forge can reduce event frequency by 40–60%, you can construct a value-based price and defend it. Without the pain quantification, ACV is a guess.
  • 1.8 · Competitive Landscape — What You Are Actually Up Against — next lesson: once you know the size and shape of the market, you need to understand who else is serving it and what the realistic competitive dynamics are. Market size without competitive context is only half the picture; the next lesson completes it.

Market sizing is the synthesis exercise of Stage 1 discovery. It takes everything you learned — ICP definition, pain quantification, signal reading, assumption mapping — and converts it into a quantitative model of the opportunity. Done well, it does not just answer "is this market big enough?" It reveals which assumption you most need to get right, which expansion path is most plausible, and which part of the model an investor will push hardest on. That is what makes it worth doing carefully.

GLOSSARYKey terms

TAM — Total Addressable Market
The maximum revenue opportunity if every potential buyer in the defined market segment purchased your product. In bottom-up sizing: total potential customer count × ACV. Used primarily to establish category magnitude; not a target revenue figure.
SAM — Serviceable Addressable Market
The portion of TAM that is reachable given your actual go-to-market motion, geographic focus, and ICP definition. The ceiling on your revenue if you converted every reachable ICP. Formula: Reachable ICP count × ACV.
SOM — Serviceable Obtainable Market
The realistic near-term revenue target, accounting for conversion rate and expansion economics. Formula: Reachable ICP × Conversion rate × ACV × Expansion multiplier. SOM should anchor your 3-year revenue model.
Bottom-up market sizing
A method that derives market size by counting reachable buyers and multiplying by a per-unit price, building from specific, defensible inputs rather than from macro industry reports. Preferred by sophisticated investors because every assumption is arguable.
Top-down market sizing
A method that starts from a macro market size (typically from an industry report) and narrows via assumed market-share percentages. Useful for establishing category magnitude; dangerous when used to derive revenue targets or investment theses.
ACV — Annual Contract Value
The average revenue per customer per year, net of one-time fees. The highest-leverage single assumption in a market size model: a 25% increase in ACV increases SAM and SOM by 25% each, without requiring more customers or higher conversion. Should be anchored to customer value delivery, not cost-plus logic.
Expansion rate
The multiplier reflecting average revenue growth per customer over time, via additional users, additional lines/facilities, or additional product modules. An expansion rate of 1.3× means the average customer is generating 30% more in year 2 than at initial close. Critical for SOM calculations in usage-based or seat-based models.
ICP count (denominator)
The number of companies or facilities that meet the Ideal Customer Profile criteria and are reachable by the company's go-to-market motion. The denominator of any bottom-up market size calculation; must be derived from primary or verifiable secondary sources, not assumed from macro statistics.

ACTIVE RECALLCheck your understanding

Answer in your own words first, then reveal.

1A founder shows you this market size: "The global predictive maintenance market is $14.2B, growing at 26% CAGR. We are targeting a conservative 0.5% share, giving us a $71M opportunity." What specific questions would reveal whether this is real analysis or theater?›

Ask: Where does the $14.2B figure come from, and does it actually describe your customer segment? (Global market reports typically aggregate adjacent markets and use broad definitions that may not match the specific ICP.) How many actual buyers fit your ICP, and where did you count them? (If they cannot answer with a specific number derived from NAICS codes, databases, or direct counting, the denominator is invented.) What would they pay per year, and how did you arrive at that ACV? (If ACV is not anchored to discovery interviews, existing spend, or value delivered, it is assumed.) What conversion rate do you need to close 0.5% of this market? (If the implied conversion rate is 20%+ of a precisely defined universe, it is not conservative — it is implausible.) Any top-down analysis that cannot answer these questions has not been stress-tested.

2In Forge's bottom-up model, ACV is the highest-leverage assumption. Why does this make undercharging specifically dangerous — not just for revenue, but for the market model and the fundraising narrative?›

Because ACV multiplies into every number in the model. If Forge charges $20K ACV instead of $32K, the SAM falls from $40M to $24M and the SOM falls proportionally. A $24M SAM is below the threshold where most venture investors see a fundable series A opportunity (they typically want 10–20× their target ownership value in SAM). So undercharging does not just reduce revenue — it reframes Forge as a lifestyle business or a company that needs to acquire 2× as many customers to reach the same milestone, which doubles the go-to-market cost, extends the runway required, and increases execution risk. Founders who undercharge to accelerate adoption are trading short-term growth optics for long-term business model fragility. The market size model makes this concrete and quantitative.

3When is top-down TAM analysis actually useful, and when does relying on it become a credibility problem?›

Useful for: establishing that the category is large enough to support a venture-scale outcome — "the industrial inspection software market is at least $8B globally, confirming this is not a niche problem." This calibrates the conversation and signals that the founder understands the macro landscape. Becomes a credibility problem when: it is used to derive actual revenue targets (the "1% of a big number" move), when the market report aggregates categories that do not match the ICP, when the share assumption is invented rather than derived from conversion rates and customer counts, or when the founder cannot answer "how many buyers is that?" with a number that comes from counting rather than a percentage. Sophisticated investors use top-down numbers as context but build their own bottom-up models to evaluate the opportunity — founders who cannot also do this have ceded the most important part of the diligence process to the investor.

4Forge's discovery found that Tier-1 plants lose $180K–$400K per unplanned downtime event. How should this translate into an ACV figure, and what would make the resulting ACV "defensible" in an investor meeting?›

The translation requires an estimate of Forge's impact on downtime event frequency. If discovery evidence suggests Forge reduces unplanned downtime events by 40–60% — derived from the detection latency improvement documented in interviews — and plants average 1–2 such events per year, then Forge's average value delivered is roughly 0.5–1.2 events prevented × $250K (midpoint) = $125K–$300K per year per facility. Pricing at $28K–$40K ACV captures 9–32% of value delivered, which is consistent with enterprise software pricing norms (typically 10–30% of documented ROI). A defensible ACV presentation connects: (1) the pain quantification from discovery ("plants in our sample reported $180K–$400K average downtime cost"), (2) the impact estimate ("our sensor reduces event frequency by 40–60% based on test deployments"), and (3) the value-capture ratio ("we price at approximately 15–25% of delivered value, consistent with comparable enterprise software ROI models"). Any ACV without this chain is an assumption, not an anchor.

Next: mapping the competitive landscape — who is actually in your market and why that changes your go-to-market →