Startup Valuation Calculator

Methodology 2026-08-v2.3 — How the Calculator Works

This tool estimates a headline equity pre-money ballpark with the comparable-transactions revenue-multiple method. Output ≈ adjusted revenue × comps median multiple. It is not a 409A, DCF, fairness opinion, or post-money figure. Optional Advanced inputs add an EV → equity net-debt bridge and a common-stock terms haircut. Data: N ≈ 2,592 private venture rounds with revenue multiples; statistic = median by industry × round (Angel / Seed / Series A / Series B / Series C+).

Pipeline in order: comps cell + shrinkage → revenue base by stage → growth γ → quality / GM / NRR → size → optional country risk → market temperature → Φ clip → optional net-debt bridge → optional terms haircut → point + range.

Last updated: 2026-08-21 · comps data as of Q3 2026 · methodology 2026-08-v2.3

Observed comps baselines (live grid)

These All-Industries medians are the fallback when an industry × round cell is thin. Sector medians can differ substantially.

Funding round	Median EV/Revenue	p25 – p75	Sample size (n)
Angel	10.0x	7.1x – 11.9x	69
Seed	12.7x	7.2x – 29.3x	251
Series A	7.6x	6.0x – 10.9x	102
Series B	5.2x	4.5x – 8.0x	62
Series C	5.0x	4.7x – 6.6x	54
Series D	5.0x	5.3x – 7.1x	45
Series E	6.6x	6.4x – 7.0x	30

Industry	Seed median	Series A median	Series B median
AI (all AI-tagged)	10.9x	12.3x	29.0x
Native AI / GenAI	12.3x	13.6x	29.0x
FinTech	22.1x	6.8x	9.3x
Enterprise Applications	24.0x	8.4x	9.3x
Enterprise Infrastructure	9.5x	3.9x	5.1x
HealthTech	12.2x	10.5x	14.2x
Consumer	16.3x	17.1x	2.8x
EdTech	20.3x	12.2x	10.0x
Food and Agriculture	13.1x	13.4x	14.3x

Source: CrackTheDeck comps database (Comparables_ext), disclosed private-market rounds. Median = 50th percentile of EV/Revenue for the industry-round cell. Open Advanced on the calculator for GM, NRR, country risk (λ), net debt, and the terms haircut.

Why Revenue Multiples

For pre-profit firms, earnings multiples break down; sales multiples stay defined. Damodaran’s stable-growth identity ties EV/Sales to margin, reinvestment, growth, and risk — every adjustment below maps to a term of that identity or to a documented empirical regularity (growth, retention, scale, country risk, market heat).

Our grid is private venture transaction comps, not public trading comps. A private-company discount is not applied on top (comps are already private rounds; adding PCD would double-count).

Step A — Comps Multiple + Shrinkage

Cell medians on small n are noisy. We shrink toward the round’s All-Industries median (Vasicek / James–Stein structure):

w = n / (n + 8) M₀ = w · M_industry + (1 − w) · M_all (null / missing n → w = 0 → full fallback to All Industries)

Announced VC rounds oversample fundable companies (survivorship). Medians describe firms that closed a priced round, not all startups — disclosed, not “fixed.” Thin cells also widen the output range.

Step B — Revenue Base by Stage

Round	Default base	Rationale
Angel, Seed	NTM revenue	Early rounds price the plan; trailing is often meaningless
Series A / B / C+	TTM revenue	Trailing comps base; optional forward blend in Advanced

There is no universal forward ×0.60 haircut (removed from v1). Guard: NTM/TTM > 10 flags an input warning. If TTM is missing or the ratio is that extreme on Series A/B/C+, the model prices the NTM base like an early round with γ = 1 (growth already in the base) and disables forward blend.

Step C — Two-Sided Growth Factor γ (Series A/B/C+ only)

g = (NTM − TTM) / TTM g ≥ 0: γ = 1 + 0.30 · min(g, 1) → premium up to ×1.30 g < 0: γ = 1 + 0.50 · max(g, −0.5) → discount down to ×0.75

Continuous at g = 0, no cliffs. Angel/Seed: γ = 1 (growth already in the NTM base).

Why flat or declining revenue is discounted, not just un-rewarded

1. Valuation identity. The sales multiple compresses as growth falls — M ∝ 1/(r − g) is monotone in g on both sides of zero. Absence of growth is a lower multiple, not a neutral one.
2. Reference-set selection. Comps medians come from companies that closed priced venture rounds — almost all growing. A flat or shrinking company deviates negatively from that reference set in a way the median cannot express; the discount restores comparability.
3. Asymmetric slope (0.50 down vs 0.30 up). Revenue declines are more persistent than growth spurts (churn compounds), and decline raises the effective discount rate (distress channel) while cutting expected margin — two channels against one. Practitioner ARR tables price declining ARR at deep discounts to stage medians.
4. Floor ×0.75 at g ≤ −50%. Deeper decline is distress pricing outside the scope of venture comps; the model refuses to extrapolate there. The overall Φ clip [0.50, 1.80] still guards the full stack — NRR < 100% and g < 0 may legitimately stack (retention of existing base vs total top line are different measures).

If NTM is not provided on Series A/B/C+, growth is unknown: γ = 1 with a caveat that the stage median already embeds typical growth of funded companies — a non-growing business likely sits below it.

Step D — Revenue Quality (software whitelist only)

Applies only to: Enterprise Applications, Enterprise Infrastructure, AI / Native AI, High Tech, Business Services, FinTech, EdTech, HealthTech.

• Recurring share s (basic): f_qual = 1 − 0.30·(1 − s)
• Gross margin (Advanced; overrides f_qual if set): ≥75% → 1.10 · 50–75% → 1.00 · 30–50% → 0.85 · <30% → 0.70
• NRR (Advanced; only if s ≥ 50%): ≥120% → 1.15 · 110–120% → 1.07 · 100–110% → 1.00 · <100% → 0.85. Premium only if growth g > 0; otherwise f_nrr = 1 + warning.

GM and recurring share are mutually exclusive proxies for margin — both together would double-count.

Step E — Size Factor (Series A/B/C+)

TTM revenue	f_size
< $20M	1.00
$20–50M	0.92
> $50M	0.85

Step F — Optional Forward Blend

If NTM is given and Advanced “forward blend” is on: mix TTM-based and NTM-based values with stage weights ω_A=0.6, ω_B=0.7, ω_C+=0.8 on realized TTM. The NTM leg uses the factor stack without γ. Default: off.

Step G1 — Country Risk (CDS-based CRP)

Geography prices into venture rounds (micro fixed-effects evidence). Macro level comes from credit markets via Damodaran’s Country Default Spreads and Risk Premiums (CDS-based), embedded as CRP_TABLE (~178 countries, as-of Jan 2026).

How CRP is built

1. Default spread = traded sovereign CDS vs default-free (US) CDS where available; else rating-implied spread (Moody’s / S&P → average CDS by rating bucket); frontier unrated markets via Damodaran’s PRS-composite mapping.
2. Country risk premium (CRP) = default spread × relative equity/bond volatility (~1.52×, 2021–25 average). Equity markets are riskier than sovereign bonds, so the credit spread is scaled up.
3. Mature-market ERP anchor ≈ 4.23% (Jan 2026); total country ERP = mature ERP + CRP.

A higher discount rate compresses the sales multiple through 1/(r − g). We convert CRP into a multiple factor with a formula, not a hand-set regional bucket:

f_ctry = (r_R − g_lt) / (r_R + λ · CRP_c − g_lt) clipped to [0.70, 1.00]

Component	Value	Notes
CRP_c	per country from CRP_TABLE	CDS / rating / PRS; shown in Advanced results with as-of date
r_R (stage discount)	Angel/Seed 50% · A 35% · B 30% · C+ 25%	Venture-practice ranges (Sahlman; Metrick–Yasuda)
g_lt	5%	Stable-growth anchor
λ (exposure)	1.0 local revenue · 0.5 HQ local / revenue global · 0.25 registration only	Risk follows where the company operates, not its passport

Worked CRP examples (Series A, r_R = 35%, λ = 1)

Country	CRP	f_ctry
United States	0.23%	≈ 0.99
Germany / Singapore	0%	1.00
United Arab Emirates	0.64%	≈ 0.98
United Kingdom / Saudi Arabia	0.78%	≈ 0.97
Israel	2.07%	≈ 0.94
India	2.85%	≈ 0.91
Brazil	3.24%	≈ 0.90
Turkey	4.66%	≈ 0.87
Nigeria	8.41%	≈ 0.78

Flat regional haircuts (e.g. “GCC ×0.80”) were too harsh for low-CRP Gulf markets and too soft for frontier names. The CDS anchor fixes the level; λ softens the penalty for globally selling companies that are only registered locally.

Coverage and fallback

• Table covers ~178 countries (rated CDS-calibrated + unrated frontier via PRS).
• Country not in table → use the worst CRP in the table (currently 26.66%, e.g. Belarus/Sudan). At that CRP the raw factor hits the 0.70 floor; the UI warns “fallback: worst-country CRP.” Never silently treat unknowns as 1.00.
• Skip country in the form → f_ctry = 1.00 (no adjustment).
• If crp_as_of is older than 12 months: point estimate unchanged; range widens + staleness note.

Step G2–G3 — Market Temperature & Hot Sectors

f_mkt ∈ [0.85, 1.15], admin-set with data vintages (default 1.0). AI / Native AI sectors get a reversal warning only (Que & Zhang, 2021) — no numeric premium for attention heat.

Step H0 — EV → Equity Bridge (optional net debt)

Revenue multiples price the operating business (EV-consistent). Equity holders own what is left after creditors:

Equity = Value_from_multiples − Net debt Net debt = interest-bearing debt − cash & equivalents

• For typical venture-backed startups net debt ≈ 0 (little debt, VC cash on the balance sheet) — the headline round valuation and EV are close, so the field is optional and blank by default.
• Enter a positive figure for venture debt / bridge notes — it is subtracted. Enter a negative figure for net cash — it adds to equity.
• The uncertainty band applies to the multiples (EV) side; the bridge shifts the band rather than scaling it. Equity is floored at $0 with a distress warning if debt exceeds the multiples-implied value; net debt above 30% of it triggers a plausibility warning.

Step H — Terms Haircut (optional)

Comps are headline round valuations. Research shows post-money headlines often exceed fair value of common (~39% average; ~48% for unicorns — Gornall & Strebulaev). Toggle “fair value of common” applies f_terms = 0.75 after the Φ clip and the net-debt bridge (it re-prices the equity claim, not the business) and relabels the output.

Step I–J — Stack Clip, Point, Range

Φ = γ · f_qual|gm · f_nrr · f_size · f_ctry · f_mkt Φ ← clip(Φ, 0.50, 1.80) V = X_ref · M₀ · Φ (× f_terms if toggled) M_eff = V / X_ref

Range uses comps IQR when available; otherwise an uncertainty band that widens on thin cells and each active optional factor (+2 pp), plus staleness widening. Confidence: High / Medium / Low from n, shrink weight w, and factor count k.

Worked Example (FinTech · Series A)

Inputs: TTM $2M, NTM $4M, recurring 100%, no advanced quality. Cell FinTech × A median 6.8 (n=75) shrinks with All-A → M₀ ≈ 6.97. g = 100% → γ = 1.30. Φ = 1.30 → V ≈ $18.1M (range ~$15.4–20.8M, High confidence).

Same + NRR 125%, GM 80%, UAE local (λ=1): f_gm=1.10, f_nrr=1.15, CRP 0.64% → f_ctry≈0.98 → Φ≈1.61 → V ≈ $22.4M. Same fundamentals in Nigeria (CRP 8.41%) → f_ctry≈0.78 → V ≈ $17.9M — country risk alone takes ~22% off the multiple with an auditable formula.

What This Is Not

• Equity view by default assumes net debt ≈ 0 — enter net debt in Advanced to apply the EV → equity bridge explicitly.
• Not post-money (new money not modeled).
• Not preference-waterfall aware beyond the flat terms toggle.
• Not selection-bias corrected — closed-round survivorship is disclosed.
• Calibrated bands (γ, GM/NRR/size/geo) are policy defaults with literature-backed sign — revisited each data vintage.

Frequently Asked Questions

What is a revenue multiple?

Company value divided by revenue (e.g. $20M value on $2M revenue = 10×). For unprofitable startups it is the practical market-approach anchor.

How do I calculate valuation from revenue?

Take the industry × stage comps median (shrunk if thin), choose TTM or NTM by stage, multiply by the factor stack above. The calculator runs that pipeline and shows each factor.

Why do seed multiples look higher than Series A?

Investors price potential against a small revenue base; as ARR scales, multiples compress even when absolute valuation rises. Angel/Seed also apply the multiple to NTM, not TTM.

How does country risk change my number?

Pick operations country and exposure λ in Advanced. We look up Damodaran CRP, compute f_ctry = (r − g) / (r + λ·CRP − g), clip to 0.70–1.00, and multiply into Φ. Skip country to leave the multiple unadjusted for geography.

Is this a substitute for a 409A valuation?

No. This calculator produces a fundraising-conversation benchmark based on private-market comps, not a formal 409A appraisal for tax purposes, not a fairness opinion, and not investment advice.

Are multiples trailing or forward?

Comps are trailing revenue multiples. Angel/Seed apply them to NTM. Series A+ default to TTM; optional blend mixes bases without a blanket −40% haircut.