Reference

Venture benchmarks investors actually use

Most benchmark pages blur two very different things: figures a named publication reports, and conventions the industry repeats. This page keeps them apart. Quote either one freely — a link back to this page is all we ask.

Sourced figures are quoted from the linked publication and dated. Rules of thumb are widely used industry conventions with no single authoritative source, and are labelled separately so nobody cites them as data.

01

Why startups fail

Failure analysis is the most useful benchmark set in diligence, because it tells you which risks actually kill companies rather than which ones are easiest to write about.

SourcedQuoted from the linked publication
Rule of thumbIndustry convention, not a measurement
  • Cause of death versus cause on the certificate

    Treat "ran out of cash" as a symptom and look one step upstream for the demand, pricing or delivery problem that consumed the cash.

    Two companies with identical burn fail for different reasons; only the upstream cause is diligenceable before the fact.

02

Unit economics

The thresholds below are the ones investors quote in committee. They are conventions, not measurements, so we label them as such — and they only mean anything once the inputs behind them are verified.

Rule of thumbIndustry convention, not a measurement
  • CAC payback

    Under 12 months is strong for SMB, under 18-24 months is acceptable for enterprise.

    Payback is the closest single number to capital efficiency, and it is computed on gross profit rather than revenue.

  • LTV to CAC

    3x or better, with LTV built from gross margin and observed retention, not list price.

    Almost every inflated LTV:CAC comes from using revenue instead of gross profit, or an assumed churn rate.

  • Net revenue retention

    Above 100% signals expansion; below 90% means growth is a treadmill.

    NRR determines whether new sales compound or merely replace churn.

  • Gross margin

    70%+ for software; materially lower means the business is partly a services or hardware business.

    Margin sets the ceiling on how much can be spent to acquire a customer.

  • Burn multiple

    Net burn divided by net new ARR: under 1.5x is efficient, above 3x needs an explanation.

    It normalises efficiency across stages better than growth rate alone.

03

Funding and dilution

Round sizes move with the cycle, so treat any single number as perishable. The structural conventions below move much more slowly.

Rule of thumbIndustry convention, not a measurement
  • Dilution per priced round

    15-25% is the normal band; a seed round is usually planned around 20%.

    Sizing the round from the milestone rather than from a valuation keeps dilution inside the band.

  • Option pool

    10-15% created or topped up at each round, typically pre-money and therefore founder-diluting.

    The pool is frequently the largest unnoticed component of a term sheet's real dilution.

  • Runway raised

    18-24 months of runway per round, with the raise starting at 9-12 months remaining.

    Fundraising takes a quarter or more; starting below six months converts a negotiation into a rescue.

  • Milestone coverage

    A round should fund the specific evidence the next investor needs, plus a margin for it arriving late.

    Rounds sized to a valuation instead of a milestone tend to end one proof point short.

04

Market sizing

There is no benchmark for a correct market size — only a correct method. These are the checks we apply to every number the platform accepts.

Rule of thumbIndustry convention, not a measurement
  • Direction of construction

    Build bottom-up from buyer counts and prices; reject any figure derived as a percentage of a headline market.

    Top-down sizing cannot be falsified, so it carries no information for a committee.

  • SOM horizon

    Derive SOM from channel capacity over a stated period, usually three years, not from an aspirational share.

    Capacity is checkable: reps, conversion rates, pipeline coverage.

  • Label every input

    Each input is verified, estimated or assumed, and the weakest label caps the confidence of the output.

    One assumed conversion rate can dominate a market size built from otherwise solid data.

Apply them

Run these thresholds against a real deal

The calculators use the same conventions listed above, and the guides explain the method behind each one.