Burn Multiple: The Efficiency Metric That Matters

The single number that ended the growth-at-all-costs era is Burn Multiple: The Efficiency Metric That Matters, and it answers a question every vanity metric dodges. How much cash did you set on fire to add one dollar of recurring revenue?

Net burn divided by net new ARR. That is the whole formula. A burn multiple of 1.0x means you spent a dollar of cash to add a dollar of annual recurring revenue. A burn multiple of 3.0x means you spent three.

A low number says the growth is paying for itself. A high one says you are renting growth from your balance sheet, and in 2023 the market stopped funding that rental.

This piece reads the metric through its canonical source, the benchmark bands that now gate fundraising, and the public filings that show why two companies with the same burn multiple are not equally efficient. The framing is analytical, not advisory: how to locate efficiency, not what to do about any stock.

Key takeaways

• The formula is one line. Burn multiple equals net burn divided by net new ARR. Coined and popularized by David Sacks of Craft Ventures in April 2020 (Sacks, “The Burn Multiple,” 2020). Lower is better.
• The bands gate term sheets. Below 1.0x is excellent, 1.0-1.5x good, 1.5-2.0x acceptable, above 2.0x problematic (Sacks framework via WallStreetPrep, Drivetrain.ai). The reported median Series A sits near 1.6x in 2025 (CFO Advisors, 2025).
• Gross margin breaks the tie. A 1.5x burn multiple on Snowflake’s 67% GAAP gross margin (FY2025 press release) is not the same business as 1.5x on Adobe’s 89.3% (Form 10-K, FY2025).
• 2023 was the turn. Capital efficiency moved from a nice-to-have to the primary filter as funding tightened (Sapphire Ventures, 2023-2024 review).
• The inverse exists too. Bessemer’s Efficiency Score is net new ARR over net burn, the reciprocal of burn multiple (Bessemer Venture Partners; MetricHQ).
• It is a snapshot, not a verdict. Burn multiple says nothing about revenue quality or churn underneath the ARR. The bear case below weighs that honestly.

---

What is burn multiple, and where did it come from?

Burn multiple is net burn divided by net new ARR. It was coined and popularized by David Sacks of Craft Ventures in April 2020 (Sacks, “The Burn Multiple,” Substack, April 23, 2020). It measures how much cash a company burns to generate each incremental dollar of annual recurring revenue, and lower is better.

The intuition is brutal in its simplicity. Most efficiency metrics ask about a slice of the business: CAC asks about sales, gross margin asks about delivery, the magic number asks about sales-and-marketing yield. Burn multiple asks about the whole machine at once. It takes every dollar that left the bank, divides by every dollar of durable new revenue that arrived, and returns a single ratio.

Sacks framed it as the metric that survives a downturn. In a market flush with capital, you can post any growth rate you want if you are willing to burn enough to buy it. Burn multiple strips that out. It does not reward the growth; it prices the cost of the growth. A company growing 100% on a 1.0x burn multiple is a fundamentally different animal from one growing 100% on a 4.0x burn multiple, even though the top-line story sounds identical.

That is the same discipline that runs through every unit-economics post on this site. The headline number flatters; the conversion rate underneath it tells the truth. The logic that makes gross margin destiny in SaaS operates one altitude up here: margin sets how much of each revenue dollar survives, burn multiple sets how much cash it took to win that dollar in the first place.

---

Why does burn multiple beat growth rate as an efficiency signal?

Because growth rate is a measure of output with no measure of input. Burn multiple puts the input in the denominator and the output in the numerator, so it cannot be gamed by spending. Two companies can post the same growth rate while one is efficient and one is lighting cash on fire; only burn multiple separates them.

Run the contrast. Growth rate tells you a company added a given amount of ARR. It says nothing about what that ARR cost. You can manufacture almost any growth rate with a large enough burn, which is exactly what the 2021 market rewarded and the 2023 market punished.

Burn multiple closes that gap by making the cost explicit. The denominator is the result you wanted (net new ARR); the numerator is what you paid for it (net burn). A founder cannot improve the ratio by spending more, because spending more raises the numerator. The only way to improve it is to make each dollar of growth cheaper, which is the actual goal.

This is why it became the filter when capital tightened. Growth rate is a fair-weather metric. It looks great until the funding that subsidized it disappears, at which point the company discovers its growth was rented. Owned, self-funding growth compounds; rented growth stops the moment the rent does.

---

The benchmark bands: reading the canonical scale

The metric is only useful with a scale, and the canonical scale comes straight from the Sacks framework as interpreted by the strategic-finance community (WallStreetPrep; Drivetrain.ai). The bands are blunt on purpose: they are a triage tool, not a precision instrument.

Below 1.0x is excellent. You are adding more ARR than cash you burn, which is the rarefied air of a genuinely efficient growth engine. From 1.0x to 1.5x is good. From 1.5x to 2.0x is acceptable but worth monitoring, the zone where you want a clear story for why each band-step exists. Above 2.0x is problematic, and the higher it climbs, the louder the question of why growth costs so much.

Burn multiple band	What it signals	Source
Below 1.0x	Excellent efficiency; growth self-funds	Sacks framework via WallStreetPrep
1.0x to 1.5x	Good; defensible in most markets	Sacks framework via WallStreetPrep
1.5x to 2.0x	Acceptable but warrants monitoring	Sacks framework via WallStreetPrep
Above 2.0x	Problematic; growth is expensive	Sacks framework via WallStreetPrep

Canonical bands per David Sacks, “The Burn Multiple” (2020), as interpreted by WallStreetPrep and Drivetrain.ai. The reported 2025 median Series A burn multiple was near 1.6x (CFO Advisors, 2025).

One reported anchor calibrates the bands against the real market. CFO Advisors put the 2025 median Series A burn multiple near 1.6x, meaning the typical company spent about $1.60 of net burn per dollar of net new ARR (CFO Advisors, “2025 Burn-Multiple Benchmarks,” 2025). That figure is a reported benchmark from a strategic-finance advisory, not an audited filing, and it should be read as a directional industry estimate rather than a precise constant.

---

The Burn Multiple Ladder

Here is the analytical asset this piece is built around. Call it the Burn Multiple Ladder: a single-screen framework that maps each band to what it signals about the business and what an operator should actually do at that rung. It is meant to be citeable on its own.

The Ladder takes the canonical bands and adds the missing column, the one founders and analysts actually need: the move. A diagnosis with no prescription is just a label. The Ladder pairs each rung with the action that band calls for.

Rung	Burn multiple range	What it signals	What to do at this rung
Top	Below 1.0x	Growth is self-funding; the engine prints durable ARR for less cash than it generates	Press the advantage. Invest into the efficient channel before competitors copy it; document why it works
Upper-middle	1.0x to 1.5x	Healthy, fundable efficiency in nearly any market	Protect it. Watch for the first sign of band-creep as you scale spend; defend the cost floor
Lower-middle	1.5x to 2.0x	Acceptable, but the engine is getting expensive	Diagnose before you raise. Find which lever slipped: pricing, churn, or cost; have the story ready for diligence
Bottom	Above 2.0x	Growth is being rented from the balance sheet	Fix the unit economics before raising more. More capital at this rung deepens the hole, it does not climb the Ladder

The Burn Multiple Ladder. Band boundaries follow the David Sacks framework (2020) via WallStreetPrep and Drivetrain.ai; the “what to do” column is original analysis.

Read the Ladder top to bottom and one rule emerges. The fix for a bad burn multiple is never simply more money. Capital raised against a 3.0x burn multiple buys more of the same expensive growth. The rung you are on tells you whether to press, protect, diagnose, or repair, and only the bottom rung is fixed by adding capital after the unit economics are sound.

---

An illustrative calculation: how the math actually runs

The formula is one line, but seeing it compute makes the levers visible. The numbers below are illustrative and hypothetical, used only to show the mechanism. They are not drawn from any filing.

Take a company that started the year at $4M ARR and ended at $7M ARR. Net new ARR is $3M. Over the same year it burned $4.5M of net cash. The burn multiple is $4.5M divided by $3M, which is 1.5x. On the Ladder, that is the lower-middle rung: acceptable, getting expensive, time to diagnose.

Illustrative (hypothetical)	Value
Starting ARR	$4.0M
Ending ARR	$7.0M
Net new ARR	$3.0M
Net cash burned	$4.5M
Burn multiple	$4.5M / $3.0M = 1.5x
Bessemer Efficiency Score (inverse)	$3.0M / $4.5M = 0.67x

Illustrative figures invented to show the calculation, not sourced to any company filing. Bessemer Efficiency Score is the reciprocal of burn multiple (Bessemer Venture Partners; MetricHQ).

Now watch the sensitivity. Hold burn at $4.5M and assume a 20% price increase that lands with no change in close rate, lifting net new ARR from $3.0M to $3.6M. The burn multiple falls to 1.25x, a full Ladder rung’s worth of improvement from a pricing decision alone. That is why the denominator is where the fastest gains usually hide, and why pricing decisions move the ratio faster than cost cuts.

Methodology: how to compute and read a burn multiple

• Inputs: net cash burned over a period (operating cash outflow plus capex, net of inflows) and net new ARR over the same period (ending ARR minus starting ARR, which already nets out churn).
• Assumptions: the period is consistent on both sides (same quarter or year), and net new ARR is genuinely net, capturing churn and contraction rather than gross new bookings.
• Sensitivity: because net new ARR is a difference of two large numbers, the ratio is volatile at small scale. A modest churn event can swing the multiple sharply, so single-quarter readings deserve less weight than trailing-twelve-month figures.
• What this misses: the metric says nothing about revenue quality, gross margin, or the durability of the ARR underneath. A clean burn multiple on low-quality, high-churn revenue is a trap, which is the whole subject of the bear case below.

---

Why gross margin breaks the tie between two equal burn multiples

Two companies with an identical 1.5x burn multiple are not equally efficient if their gross margins differ. The one with higher gross margin keeps more of each revenue dollar to cover operating expense, so its growth rests on a stronger base. Burn multiple measures cash efficiency; it does not see the cost structure beneath the ARR.

This is the trap in reading burn multiple in isolation. Consider two real, public cost structures as anchors. Snowflake reported a 67% GAAP gross margin for FY2025 (Snowflake press release, February 26, 2025). Adobe reported an 89.3% gross margin for FY2025 (Adobe Form 10-K, fiscal year ended November 28, 2025). That is a structural gap of more than 22 points.

Now imagine both posted the same 1.5x burn multiple. The Adobe-shaped business keeps roughly 89 cents of every revenue dollar after cost of revenue; the Snowflake-shaped business keeps about 67. The high-margin company’s ARR throws off far more cash to fund the next dollar of growth. Same burn multiple, very different forward efficiency, because one company’s recurring revenue is structurally more spendable than the other’s.

The lesson is to read burn multiple and gross margin together. A low burn multiple on a thin-margin base is more fragile than the ratio alone suggests, because the revenue it generates funds less of the next cycle. This is precisely why gross margin is destiny in SaaS: margin sets how much of each ARR dollar survives to do work, and burn multiple inherits that base.

---

Burn multiple vs. Rule of 40: which one moves valuation

In a tight-capital market, burn multiple beats both raw growth rate and even the Rule of 40, because it isolates the one thing scarce capital cares about: efficiency. The Rule of 40 blends growth and profitability into a single threshold, which is useful but lets a company pass on growth alone. Burn multiple cannot be passed on growth alone.

The Rule of 40 says growth rate plus profit margin should clear 40. It is a fine health check, and Rule of 40, read for what it actually tells you, remains a standard screen. But it has a blind spot: a company growing 60% with a negative-20 margin clears 40, even if it is burning enormous cash to buy that growth. The blend hides the burn.

Burn multiple closes that blind spot. It does not let growth paper over inefficiency, because the burn is the denominator’s twin in the numerator. In a market where capital is cheap, the Rule of 40’s tolerance for burn-funded growth is fine. In a market where capital is expensive, that tolerance is the exact thing investors stopped funding, which is why reported analyses now frame burn multiple as the metric that moves SaaS valuation (Bookman Capital, 2025).

The honest framing is that they answer different questions. Rule of 40 asks “is this a balanced, healthy business?” Burn multiple asks “how expensive is this company’s growth?” When capital is scarce, the second question dominates the first. The two are complements, not substitutes, and the CAC-side view of the same efficiency question lives in why CAC payback period is the SaaS metric that matters.

---

The AI-native advantage in burn multiple

Reported 2025 benchmarks show AI-native SaaS companies clustering under 1.5x burn multiples while the traditional Series A median sits near 1.6x (CFO Advisors, 2025). The mechanism is that AI compresses operating cost without compressing growth, improving both sides of the ratio at once. The advantage is reported, not audited, and the gap is early.

Walk the two sides of the fraction. On the numerator, AI lets a smaller team ship product faster, automates large parts of customer support, and makes go-to-market workflows more efficient, which holds net burn down. On the denominator, the same capabilities can accelerate net new ARR by getting product to market and customers to value faster. When both move the right way, the ratio improves more than either lever would alone.

That said, this is the most fragile claim in the piece, and it deserves a flag. The reported sub-1.5x figure is a benchmark from strategic-finance sources, not a line in an audited filing. AI-native companies are young, their cohorts are small, and a favorable burn multiple in the first year can mask retention questions that only appear in year three. The advantage is real in the reported data, but it is a snapshot of a young cohort, not a proven durable structure.

The reason the advantage is plausible is structural, not hype. AI shifts work from headcount to compute, and compute scales differently from salaries. The companies positioned best are the ones with a real cost-structure edge rather than a thin wrapper over a rented model, the distinction that decides whether the efficiency holds. The cost discipline behind it is the same one mapped in the AI infrastructure market map: the layer that controls its own cost floor keeps the most of every dollar.

---

The bear case: when a clean burn multiple lies

The strongest argument against burn multiple is not that the formula is wrong. It is that the formula is blind to the quality of the ARR in its denominator, and a beautiful ratio can sit on top of a business that is quietly falling apart.

Start with churn. Burn multiple uses net new ARR, which already nets out churn, so a company can show a respectable multiple while its gross retention erodes. The net figure can look fine in a single period because new bookings mask the leak. But high churn means the company is refilling a bucket with a hole in it, and the burn multiple does not show the hole. The reason gross and net diverge, and why the gap matters, is the whole point of gross retention vs net retention in SaaS IPOs.

Next, revenue quality. Net new ARR treats every dollar as equal, but a dollar of expansion from a delighted account is worth far more than a dollar from a discounted, soon-to-churn logo. A company can juice net new ARR with aggressive discounting that posts a clean burn multiple this year and a wave of churn next year. The metric cannot tell durable revenue from rented revenue.

Then the small-scale volatility. Because net new ARR is a difference between two large numbers, the ratio is unstable for early companies. One large churned account or one delayed deal can swing a quarterly burn multiple by a full band, which makes single-period readings noisy and easy to misuse.

And the timing objection, which is the fairest. A deliberately high burn multiple can be the correct sequence for a company building a moat in a large, fast-forming market, the “spend now, the position compounds later” logic. Read in one snapshot, that investment looks like inefficiency. Read across the arc, it can be the right call.

The honest weighing: each objection is a reason to refuse to read burn multiple alone, not a reason to discard it. Pair it with gross retention, gross margin, and revenue quality and it remains the cleanest single read on whether growth is paying for itself. The bear case bounds the metric. It does not retire it.

---

Where this is vulnerable: the liquidity blind spot

A credible metric names where it breaks, and burn multiple’s most dangerous failure is that a “good” reading can mask imminent liquidity danger. The ratio measures efficiency, not survival. A company can post an excellent burn multiple and still run out of cash.

The trap is the difference between efficiency and runway. Burn multiple tells you how efficiently you convert cash into ARR. It says nothing about how much cash you have left or how long it lasts. A company at a 1.0x burn multiple with two months of runway is in far more danger than a company at 2.0x with three years of runway, even though the first reads as far more efficient.

This is why burn multiple must be read alongside two other numbers the metric itself omits: absolute net burn (the dollars leaving the bank each month) and months of runway (cash on hand divided by net burn). A clean ratio on a tiny, fast-depleting balance sheet is a warning, not a comfort.

There is also a denominator-collapse risk. The whole metric depends on net new ARR being positive and meaningful. For a company near flat or shrinking, net new ARR approaches zero, and the burn multiple spikes toward infinity or goes negative, at which point the ratio stops being interpretable at all. The metric is built for growing companies; it degrades exactly when a company most needs a clear signal. Reading these numbers straight from the source, rather than the pitch, is the discipline at the heart of how to read a tech S-1 like an operator.

---

What operators should take from this

Burn multiple does not tell you to grow or to cut. It tells you the price of the growth you already have, and whether that price is one your business can sustain without renting the difference from your balance sheet. Here is how to act on it.

1. Compute it on a trailing-twelve-month basis, not a single quarter. Net new ARR is a difference of large numbers and swings hard at small scale. The annual figure is the signal; the quarterly figure is mostly noise.
2. Attack the denominator before the numerator. Pricing and retention move net new ARR fast and cheap. A price increase that holds close rate improves the ratio immediately, and protecting a churned dollar is worth as much as adding a new one. Cost cuts come second.
3. Read it next to gross margin and runway, never alone. A clean burn multiple on a thin-margin base funds less of the next cycle, and a clean ratio on two months of cash is a liquidity warning, not a win. The metric is a slice, not the whole picture.
4. Use the Burn Multiple Ladder to set the move, not just the diagnosis. Below 1.0x, press the advantage. In the middle bands, protect and diagnose. Above 2.0x, fix the unit economics before raising, because more capital deepens an inefficient hole rather than climbing out of it.
5. Separate timing risk from structure risk. A deliberately high burn multiple to win a forming category is a strategy; a high burn multiple because nobody owns the cost floor is a problem. Know which one you are running before you defend it.
6. Distrust the AI-native benchmark until it is your own data. The reported sub-1.5x AI advantage is a young cohort’s snapshot. If you run an AI product, measure your real ratio against your real retention, not the industry headline.

As a small, illustrative analog, running an AI feature inside a product like PDF9to5 shows the lever in miniature. Routing every request to a premium rented model raises net burn with no help to ARR, pushing the burn multiple up. Caching and routing the easy majority to a cheaper path cuts that cost on identical revenue, which pulls the same ratio down. The product does not change. Only the efficiency of the dollar behind it does, which is the entire point of the metric.

---

Analysis, not investment advice. Figures are drawn from the public sources cited inline: David Sacks, “The Burn Multiple” (2020); Snowflake’s FY2025 results press release (February 26, 2025); Adobe’s Form 10-K (FY2025); and reported benchmarks from CFO Advisors, Bessemer Venture Partners, and Sapphire Ventures, labeled as reported industry estimates rather than audited filings. Frameworks here, including the Burn Multiple Ladder, are for understanding business structure and tradeoffs, not for making buy or sell decisions.

Want the full toolkit for reading filings like this, the burn-multiple worksheet, the efficiency-band scorecard, and the Burn Multiple Ladder used above? It’s in the Tech Business Analysis Playbook.