Investors read traction as evidence: growth measured as a compounded rate on revenue or active users, retention shown by cohort, and unit economics built on profit rather than revenue [1][2][3]. Market size should be built bottom-up from customers and prices, with top-down figures used only as a sanity check, because the assumptions can be tested [5]. Every metric needs one stated definition used the same way everywhere; in an October 2026 action the SEC alleged that a startup raised about $6.7 million partly by counting free demo accounts and discounted pilots as revenue-generating customers [1][7].
In this lesson you will
- Choose the handful of metrics an investor will judge your business model on.
- Present growth, retention and cohorts in ways that survive diligence.
- Explain CAC, LTV, gross margin, net burn and the burn multiple, and what each reveals.
- Build a bottom-up market size with stated, testable assumptions.
- Define metrics once and apply them consistently across the deck, model and data room.
Traction is evidence, not decoration
Traction is the part of a pitch where claims become observations. Paul Graham's long-standing advice to Y Combinator founders is that the best thing to measure growth on is revenue, and the next best, for startups not yet charging, is active users [3]. He put benchmarks on it: a good growth rate during YC is 5% to 7% a week, 10% a week is exceptional, and 1% a week suggests the founders have not yet figured out what they are doing [3]. The arithmetic explains why investors care: a company growing 1% a week grows about 1.7x in a year, while one growing 5% a week grows about 12.6x [3]. Those weekly figures describe very early companies in an accelerator, not a later-stage business, but the principle holds at every stage: pick the number that best reflects customers paying for value, and show its rate of change honestly.
Lead with the metrics your model is judged on
Different business models are underwritten on different numbers, and a deck that leads with the wrong ones signals that the founder does not know how investors will evaluate the company. Andreessen Horowitz partners noted in 2015 that people may use different definitions of the same metric, which makes it hard to read a business's health, so the starting point is knowing which metric matters and what it includes [1].
- Subscription software: annual recurring revenue (ARR) should cover only recurring components, excluding one-time and professional services fees, and bookings (contract value) should be kept separate from recognized revenue [1]. Investors value product revenue more highly than services revenue, which is non-recurring and lower margin [1].
- Marketplaces: gross merchandise value (GMV) is not revenue; revenue is the share of GMV the marketplace keeps [1]. Pear VC lists presenting GMV instead of take-rate revenue as a common market-sizing mistake as well [5].
- Consumer and social products: define 'active' explicitly, because companies use almost unlimited definitions, and show engagement ratios such as daily to monthly users [1][2].
- AI-heavy products: gross margin deserves its own slide. a16z partners reported in 2020 that AI companies often showed gross margins of 50% to 60%, below the 60% to 80%+ benchmark for comparable SaaS, partly because of cloud compute and human review costs [6].
- Fund managers: the 'traction' is the track record. For SEC-registered advisers, the Marketing Rule bars showing gross performance unless net performance appears with at least equal prominence, and bars extracted performance unless the total portfolio's results are provided or offered [9].
Growth, retention and cohorts
How growth is computed matters. Monthly growth is often shown as a simple average of monthly rates, but investors often prefer the compounded monthly growth rate (CMGR), which will be smaller than the simple average in a growing business [1]. Cumulative charts always go up and to the right for any business with activity, so they are not a valid measure of growth [1]. Retention is best shown by cohort: groups of customers who started in the same period, tracked over time [2]. a16z partners look for retention in each cohort to stabilize after a period such as 6 or 12 months, and for newer cohorts to perform better than older ones [2]. For revenue, they distinguish gross churn, which estimates the actual loss to the business, from net revenue churn, which nets out upsells and so understates losses [1]. Showing both is a credibility signal; showing only the flattering one invites the diligence question you least want.
Unit economics and capital efficiency
Unit economics answer whether growth creates value or merely consumes cash. Customer acquisition cost (CAC) should be shown on a paid basis as well as blended, because blended CAC does not show whether paid campaigns are working and profitable [1]. Lifetime value (LTV) is the present value of the future net profit from a customer, not the revenue, and gross profit should include all costs of manufacturing, delivering and supporting the product [1]. Burn rate is the rate at which cash decreases, and investors focus on net burn to see how long the money in the bank will last [1].
David Sacks's burn multiple, net burn divided by net new ARR, condenses these into one efficiency number: the higher it is, the more the company burns for each unit of growth [4]. In his 2020 essay, a 2x multiple is described as reasonable for an early-stage startup and 5x as terrible, with higher multiples tolerated while sales are incipient and expected to fall as the company matures [4]. He calls it a catch-all metric because problems such as weak gross margin or high churn eventually show up either as more burn or less net new ARR [4]. The metric suits recurring-revenue businesses; other models need an equivalent ratio of cash consumed to durable growth added.
Market sizing: bottom-up first
Pear VC's market sizing guide, based in part on a survey of 30 VCs, advises estimating market size bottom-up rather than top-down, and using top-down figures only to sanity-check the magnitude of the bottom-up estimate [5]. VCs prefer bottom-up because its assumptions can be tested and validated, while the share assumption in a top-down estimate is often an unsubstantiated afterthought [5]. Surveyed VCs most commonly wanted market size estimated five years into the future [5]. The guide flags common mistakes: assuming a 10% market share when technology companies at IPO typically held 0.1% to 2%, citing global markets that are rarely relevant, presenting non-annual revenue, and showing TAM, SAM and SOM figures without their assumptions [5]. Because there is no single definition of market size, the guide recommends stating your methodology explicitly [5].
In practice, a defensible bottom-up size is a short chain: the number of customers who have the problem and can buy, times what each would plausibly pay per year, narrowed to the segment you can reach with your current product and channel. Each link should point to evidence, such as a customer count from a public registry or your own pricing data, so an investor can change one assumption and see the result.
Define once, report consistently
The quickest way to lose an investor in diligence is for a number in the deck to differ from the same number in the model or data room. A metric definition is a small accounting policy: what counts as a customer, when revenue is recognized, which costs sit in gross margin, and how churn is measured. Write those definitions down, put them in a footnote or appendix, and use them every time, including in investor updates after the round. The stakes are real: in a litigation release dated October 7, 2026, the SEC said it had charged Kalder Inc. and its founder with raising nearly $6.7 million through false statements, alleging that the company kept two sets of books, showed investors inflated revenue and growth, and overstated customers with revenue-generating contracts when many had signed up for free demo accounts or short-term discounted pilots [7]. The founder agreed to a settlement, subject to court approval, without admitting or denying the allegations [7]. The Pitching Truthfully lesson in this course covers the antifraud rules in detail.
What's different in 2026
The Q3 2026 PitchBook-NVCA Venture Monitor, with data as of September 30, 2026, shows a market where evidence of traction is rewarded unevenly [8]. AI startups received 82.7% of US VC deal value in 2026 through Q3, the highest annual share in PitchBook's dataset, and deals of $100 million or more, about 5% of deal count, accounted for 75% of Q3 deal value [8]. The median AI valuation step-up between rounds reached 2.2x, the highest since ChatGPT launched in November 2022 [8]. J.P. Morgan's commentary in the report describes a barbell approach among VCs, keeping a broad appetite for seed deals while doubling down as proof points and annual recurring revenue materialize at later stages, and notes that unit economics sensitive to GPU availability and pricing remain a concern [8]. For fund managers, an EisnerAmper audit partner writing in the same report says DPI has eclipsed IRR as the metric of choice and that GPs should expect questions, and verification, about who sourced, led and held a board seat on historical deals [8]. Refresh these figures each quarter; the underlying lesson, that investors pay for verified, repeatable proof, does not change.
Worked example: rebuilding a metrics slide (hypothetical)
- Before: 'ARR $2.4M, 40 customers, 15% monthly growth, $50B TAM.' ARR includes a $300K implementation fee, 9 of the 40 'customers' are unpaid pilots, growth is a simple average of three unusually strong months, and the TAM is a global industry figure.
- After, revenue: recurring ARR $2.1M, with services shown separately; 31 paying customers and 9 pilots listed as pipeline, with the definition of 'customer' in a footnote.
- After, growth: 9.2% CMGR over the trailing 12 months, plus a cohort chart showing net revenue retention by quarterly cohort and gross churn shown alongside it.
- After, efficiency: gross margin including hosting and support staff; paid CAC and blended CAC; burn multiple of 1.8x for the last two quarters.
- After, market: about 14,000 U.S. firms in the target segment times an estimated $25,000 average annual contract, roughly $350M, with each input sourced and a top-down industry figure shown only as a cross-check.
Key terms
- Annual recurring revenue (ARR)
- The annualized value of recurring revenue components, excluding one-time and professional services fees.
- Compounded monthly growth rate (CMGR)
- The constant monthly rate that would take a metric from its starting to its ending value; lower than a simple average of monthly rates in a growing business.
- Cohort analysis
- Tracking groups of customers who started in the same period to see how their retention and spending change over time.
- Gross vs. net revenue churn
- Gross churn measures revenue lost from existing customers; net churn offsets that loss with upsells and so can hide it.
- Customer acquisition cost (CAC)
- The cost to acquire a customer, shown on a paid basis and a blended basis that includes organic customers.
- Lifetime value (LTV)
- The present value of the future net profit from a customer over the relationship, not the revenue.
- Burn multiple
- Net burn divided by net new ARR; a measure of how much cash a company consumes per dollar of recurring growth.
- Bottom-up market sizing
- Estimating market size from the number of reachable customers times what each would pay, with each assumption stated and testable.
Check your understanding
Sources
- [1]Andreessen Horowitz (Jeff Jordan, Anu Hariharan, Frank Chen and Preethi Kasireddy), “16 Startup Metrics,” August 21, 2015.
- [2]Andreessen Horowitz (Anu Hariharan, Frank Chen and Jeff Jordan), “16 More Startup Metrics,” September 23, 2015.
- [3]Paul Graham, “Startup = Growth,” September 2012.
- [4]David Sacks, “The Burn Multiple,” April 23, 2020.
- [5]Pear VC, “Market Sizing Guide,” August 3, 2021.
- [6]Andreessen Horowitz (Martin Casado and Matt Bornstein), “The New Business of AI (and How It's Different From Traditional Software),” February 16, 2020.
- [7]U.S. Securities and Exchange Commission, “Litigation Release No. 26667: SEC v. Kalder Inc. and Gökçe Güven,” October 7, 2026.
- [8]PitchBook and National Venture Capital Association, “Q3 2026 PitchBook-NVCA Venture Monitor,” October 2026 (data as of September 30, 2026).
- [9]Legal Information Institute, Cornell Law School, “17 CFR § 275.206(4)-1 - Investment adviser marketing,” Accessed November 2026.
Educational content only — not legal, tax, or investment advice, or an offer to sell or a solicitation of an offer to buy any security. Rules and data are described as of November 2026 and may change. Securities offered through Umergence LLC, Member FINRA/SIPC.