A Series B product-led company built its plan around one number: "normal" CAC payback for B2B SaaS. The number was twelve months. It came from a widely shared benchmark report, so it felt safe.
Eighteen months later, the company missed plan by 40%.
The benchmark was not fake. It was just the wrong comparison. It mixed companies with different ACVs, sales motions, stages, geographies, and channels. A more careful read of SaaS CFO data put median CAC payback around fifteen months across 939 B2B SaaS companies, with SMB often at 8-12 months, mid-market at 14-18, and enterprise at 18-24.
The company did not choose the wrong plan. It measured itself against the wrong companies.
The problem
Benchmarks are useful only when the cohort matches.
Most benchmark reports hide the cohort. They show a clean median and a logo. They rarely show the details that determine whether the number applies to your company.
That is why benchmarks can be dangerous. They look objective. They travel easily in decks. Boards ask for them. Operators use them to set targets. But if the cohort is wrong, the target is wrong.
More data does not fix this. A thousand companies that do not look like yours are less useful than thirty companies that do.
Why benchmarks are so persuasive
Benchmarks travel well because they reduce uncertainty to one number.
A board can ask for CAC payback. A founder can ask for ROAS. A CFO can ask for gross margin. An operator can search for a median. The number feels external, objective, and safer than judgment. It is easier to defend a plan by saying "the benchmark says" than by explaining why the company's motion is different.
That is exactly why bad benchmarks are dangerous. They do not look like opinions. They look like evidence.
The risk is not only setting the wrong target. It is learning the wrong lesson. A brand may think its paid social is underperforming when it is being compared to brands with higher AOV and better repeat purchase. A SaaS company may think its payback is too slow when it sells enterprise contracts with longer buying cycles. A marketplace may optimize for conversion rate when liquidity is the actual constraint.
Benchmarks should widen the question. They should not close it too early.
The five things a benchmark must match
A benchmark should match your company on five dimensions.
Business model. PLG, sales-led SaaS, marketplace, services, ecommerce, and subscription commerce have different economics. The same CAC payback number cannot describe all of them.
Stage. A $500k ARR company, a $10M ARR company, and a $100M ARR company should not be judged by the same rule. The Rule of 40 was built for companies at scale, not a Series A searching for repeatability.
Geography. U.S., Europe, and global cohorts face different currencies, consent rules, media prices, and buyer behavior.
Channel mix. A company driven by paid search has different payback from one driven by outbound, referrals, affiliates, SEO, influencers, retail, or partnerships.
Measurement protocol. Metrics with the same name often use different formulas. CAC payback can be gross or margin-adjusted, fully loaded or not, cohort-based or period-based. NRR, churn, ARR, MER, ROAS, and LTV all have similar definition traps.
If a benchmark does not publish these five dimensions, treat it as a rough story, not a planning number.
What a useful benchmark looks like
A useful benchmark is less flashy and more specific.
Example:
"Cohort: 47 North American B2B SaaS companies, $5M-$20M ARR, mid-market ICP, paid plus content above 50% of pipeline, measured Q3-Q4 2025. CAC payback is gross-margin-adjusted, fully loaded sales and marketing, cohort-based, six-month average. Median: 14 months. Interquartile range: 9 to 22."
That sentence is not as viral as "SaaS median CAC payback is 12 months." It is much more useful.
Good benchmark copy should tell you who is included, who is excluded, how the metric is calculated, and how wide the spread is.
The internal benchmark is usually better
Before looking outside, a company should build its own comparison set.
For an e-commerce brand, that might mean comparing:
- Meta prospecting campaigns by offer type
- landing pages by traffic source
- email flows by customer segment
- bundles by margin and repeat purchase
- affiliates by refund rate and payout terms
- channels by CAC payback and contribution margin
For a SaaS company, it might mean comparing segments by ACV, sales motion, implementation effort, support load, churn, and expansion.
Internal benchmarks are not perfect. They are smaller and can be biased by the company's own history. But they have one major advantage: the metric definitions, customers, products, and operating context are visible. A small internal comparison that the team understands often beats a large external benchmark whose cohort is vague.
How to use benchmarks without getting fooled
Before a benchmark enters a plan, ask:
- What exact companies produced this number?
- Which companies were excluded?
- How was the metric calculated?
- Does the cohort match our model, stage, geography, and channel mix?
- What is the variance, not just the median?
If the report cannot answer those questions, do not build the plan around it.
What Lyberty does
Lyberty keeps metric definitions and cohorts attached to the number.
A metric should know how it was computed, which time window it covers, which campaign or channel it applies to, and what comparison group it can safely be compared with. If two cohorts do not match, the system should say that instead of producing a confident benchmark.
That matters for decisions like:
- should we increase spend on this channel?
- is CAC payback actually worse than peers?
- is this offer underperforming or just being compared to the wrong cohort?
- is this agency report using the same revenue definition as finance?
Small, honest comparisons beat large, fuzzy ones.
What to do now
- Audit every benchmark in your plan.
- Write the cohort sentence the publisher would need to provide.
- Remove any number whose cohort sentence you cannot write.
- Build internal benchmarks first.
- Use external benchmarks as directional checks, not targets.
The goal is not more confident planning. It is more honest planning.
Sources
- Open Science Collaboration. "Estimating the reproducibility of psychological science." Science, 2015. https://www.science.org/doi/10.1126/science.aac4716
- Brad Feld. "The Rule of 40% For a Healthy SaaS Company." 2015. https://feld.com/archives/2015/02/rule-40-healthy-saas-company/
- Tomasz Tunguz. Public commentary on the Rule of 40, 2015. https://x.com/ttunguz/status/566339479858733056
- ICONIQ Growth. "The 2024 ICONIQ Growth Resiliency Rubric." https://www.iconiqcapital.com/growth/insights/iconiq-growth-resiliency-rubric
- High Alpha and OpenView. "2024 SaaS Benchmarks Report." https://www.highalpha.com/saas-benchmarks/2024
- Bessemer Venture Partners. "The BVP Nasdaq Emerging Cloud Index." https://cloudindex.bvp.com/
- Scale Venture Partners. "SaaS Metrics: A History of the Magic Number." https://www.scalevp.com/insights/saas-metrics-a-history-of-the-magic-number/
- Scale Venture Partners. "Rule of 40 Does Not Compute for Early-Stage Startups." https://www.scalevp.com/insights/rule-of-40-does-not-compute-for-early-stage-startups/
- Jeff Sauro. "Has the Net Promoter Score Been Discredited in the Academic Literature?" MeasuringU, 2017. https://measuringu.com/nps-discredited/
- The SaaS CFO. "CAC Payback Period benchmarks across 939 B2B SaaS companies." 2025. https://www.thesaascfo.com/cac-payback-period/