LeanScale · Reference Corpus · August 2026

Nobody argues about the benchmark. They argue about the denominator.

Funnel conversion, sales cycle, retention and efficiency benchmarks for B2B SaaS — every figure carrying its source, sample size and panel, and every metric carrying the definition it was actually measured on. Built so that when someone asks "are we in the fairway?", you can tell whether you're even comparing the same number.

7Primary sources
17Metrics defined
5Reconciled conflicts
3,000+Companies behind the figures
The premise

Two credible sources put B2B SaaS win rates at 43% and 19% in the same year. Both are right. They are measuring different denominators — and almost every benchmark fight is that fight, unnamed.

Start here

Five checks before a benchmark means anything.

A B2B SaaS benchmark is only comparable when the denominator, the panel, the statistic and the time basis all match. Run these five checks in order — if a comparison fails any one of them, the gap you are looking at is measurement, not performance. Most do fail, usually on the first or the third.

Match the denominator

The same metric name covers genuinely different formulas. Win rate, NRR, CAC payback and conversion rate each have three or more live conventions. Settle the formula before the number.

Match the panel

ARR band, ACV band, funding type, GTM motion, geography. A $60K-ACV enterprise seller and a $600-ACV self-serve product share almost no structural DNA. ACV is the strongest single cut.

Match the statistic

Median, mean, upper quartile. The single most common benchmarking error in the wild is comparing your median against somebody's published top quartile and concluding you're broken.

Match the time basis

Cohort (records that entered in the period, tracked to outcome) versus snapshot (records sitting in the stage now). These produce different numbers from identical data.

Check the cell, not the headline

Sample size collapses when a survey cuts by ACV. A report with n=150 can have n=9 in the band you care about. Read the per-chart n before you quote the per-report n.

So what

Benchmarks answer "are we structurally normal?" They do not answer "what should we fix?" For the second question, the only useful comparison is a company against its own trailing cohorts — which is also the only comparison where the definition is guaranteed to be constant.

The definition layer

Seventeen metrics, and the ways each one gets measured differently.

This is the part of the corpus that does the work. For each metric: the canonical formula, the variants that circulate, what each variant does to the resulting number, and which published source uses which convention. Risk pills flag how often the metric is the actual cause of a mismatch.

Funnel & pipeline metrics

Win rate

Highest mismatch risk a.k.a. close rate, opportunity win rate

Most common published formula closed-won ÷ (closed-won + closed-lost)  — measured on opportunities that reached a terminal outcome in the period.

Variants that change the number

  • Decision-only denominator. Only won + lost. Every stalled, no-decision and indefinitely-slipped deal is excluded.
  • Created-cohort denominator. All opportunities created in a period, tracked to outcome. No-decision counts as a loss.
  • Qualified-only denominator. Starts at SQL or stage 2+, discarding early-stage disqualification.
  • Dollar-weighted vs logo-weighted. Weighted by ACV rather than deal count; diverges sharply when big deals lose more often.
  • Snapshot vs cohort. Deals closed in the period vs deals created in the period.

What it does to the number

The decision-only denominator is the single largest inflator in B2B SaaS benchmarking. In enterprise, no-decision is frequently the largest loss category — excluding it can roughly double the reported rate. This is most of the distance between the 43% and the 19% in the banner above.

Source conventionsICONIQ states it explicitly in a footnote: (# closed-won ÷ (# closed-won + lost)) × 100, and reports 43% average for sales-sourced opportunities (2026). The Bridge Group reports a 19% median win rate (2024) on a broader denominator. Neither is wrong; they are not comparable.

"Does your win rate denominator include deals that went nowhere?"

Stage conversion rate

Highest mismatch risk

Canonical (cohort basis) records that advanced past stage N ÷ records that ENTERED stage N in the period  — tracked forward to outcome, not counted where they sit today.

Variants that change the number

  • Snapshot basis. Denominator is records in the stage during the period, including ones that entered long ago. This is the CRM default and it is wrong.
  • Skipped stages. Most CRMs let a deal jump stages. A skipped stage silently records zero entries and inflates that stage's conversion toward 100%.
  • Open-deal treatment. Are still-open records excluded, counted as failures, or left in the denominator? Excluding them inflates; counting them as failures deflates.
  • Stage-to-stage vs stage-to-close. "SQL conversion" may mean SQL→next stage or SQL→closed-won.
  • Immature cohorts. A cohort younger than one sales cycle has not had time to convert.

What it does to the number

Snapshot conversion is biased optimistic, because the deals that never advance accumulate in the stage and are counted as denominator only once while winners recycle through. Immature cohorts bias pessimistic. Together they make period-over-period comparison on an unstated basis close to meaningless.

Source conventions — Survey-based sources (ICONIQ, Benchmarkit) ask respondents for a rate and do not enforce a basis, so the published averages blend both conventions. This is a structural limitation of every funnel benchmark in this corpus and the reason the funnel section carries wider error bars than the retention section.

"Is that cohort-based or a snapshot? And what happens to open deals?"

MQL, SAL, SQL

Highest mismatch risk marketing / sales-accepted / sales-qualified lead

There is no industry definition These are locally defined thresholds, not standard objects. An MQL is whatever score or action a given company decided counts.

What actually varies

  • MQL threshold. A score cutoff set internally. Lowering it raises MQL volume and lowers MQL→SQL conversion, with zero change in the business.
  • SAL vs SQL. SAL = sales accepted (a routing acknowledgement). SQL = sales qualified (a judgement after contact). Many companies run only one of the two, so "MQL→SQL" in one dataset is "MQL→SAL" in another.
  • Person vs account. Lead-level counts inflate relative to account-level counts wherever buying committees are large.
  • Recycling. Does a lead that returns count as a new MQL?

What it does to the number

The MQL→SQL rate is primarily a measure of how loose your MQL threshold is, and only secondarily a measure of lead quality. It is the least portable metric in this corpus. Benchmark it against your own history, not against anyone else's panel.

Source conventionsBenchmarkit publishes MQL→SAL at a 20% median. ICONIQ publishes New Lead→MQL at 28% and MQL→SQL at 30% (2026). The stage names do not refer to the same objects across the two panels.

"What has to be true for a record to become an MQL here?"

Sales cycle length

Medium mismatch risk

Canonical close date − opportunity created date, median, measured on won deals only.

Variants that change the number

  • Start point. First touch, MQL, opportunity creation, first meeting, or qualification. First-touch-to-close can be 2–3× opportunity-to-close.
  • End point. Verbal, signature, countersignature, or first payment.
  • Won-only vs won+lost. Including losses usually shortens the number, since losses die early — but including no-decisions lengthens it dramatically.
  • Mean vs median. Cycle-time distributions have a long right tail, so the mean runs meaningfully above the median.
  • Business vs calendar days.

What it does to the number

Start-point choice is the dominant term. Two companies with identical processes can publish cycle lengths a factor of two apart purely on where they start the clock. Always convert to opportunity-created → closed-won, median before comparing.

Source conventionsICONIQ reports average weeks and does not publish a start-point definition, so its by-ACV curve should be read as a shape rather than as absolute values. The Bridge Group reports medians on AE-owned opportunities.

"Does the clock start at opportunity creation or at first touch?"

Pipeline coverage

Medium mismatch risk

Canonical open pipeline for the period ÷ quota (or target) for the period, measured at period start.

Variants that change the number

  • Stage inclusion. All open pipeline vs qualified-and-later only. This single choice routinely moves coverage by 1.5–2×.
  • Measurement moment. Snapshot at period start vs rolling average through the period.
  • Raw vs weighted. Probability-weighted pipeline against an unweighted target is a category error, but common.
  • Close date in-period only vs all open pipeline regardless of expected close.
  • Whose quota. New-logo quota, total quota including renewals, or company target.

What it does to the number

The classic "3× coverage" rule is only meaningful with the stage set attached. The right coverage number is simply 1 ÷ your own stage-to-close conversion rate, which means a company with a genuine 25% conversion needs 4×, and one at 33% needs 3×. Borrowing the rule without the conversion rate behind it is guesswork.

Source conventionsBenchmarkit currently recommends 4:1 against a historical 3:1, and notes coverage requirements correlate strongly with ACV.

"Which stages count as pipeline in that ratio?"

Quota attainment

Medium mismatch risk

Canonical reps at or above 100% of quota ÷ total reps, over a full fiscal year.

Variants that change the number

  • Ramped-only vs all reps. Excluding ramping reps raises the figure substantially at any company that is hiring.
  • Annual vs quarterly. Quarterly attainment is lower and noisier.
  • Point-in-time roster vs everyone who held the seat. Counting only reps still employed at year end removes the ones who missed and left.
  • Attainment of quota vs attainment of a prorated quota.

What it does to the number

Ramped-only plus survivor-only can add 10–15 points against an all-reps, all-holders basis. This explains most of the gap between the two figures cited opposite.

Source conventionsICONIQ reports the share of ramped AEs achieving quota: 62% for 2025. The Bridge Group reports 48% for 2026 on a broader rep base. Both are credible; the ICONIQ figure is the more flattering basis.

"Ramped reps only, or everyone who carried a number?"

ACV, ASP and ARR per customer

Highest mismatch risk

Three different metrics, routinely used interchangeably ACV = total contract value ÷ contract years  ·  ASP = average value of a closed deal  ·  ARR/customer = total ARR ÷ customer count

Why they diverge

  • Multi-year contracts. A three-year $300K TCV deal is $100K ACV — but it may bill $150K in year one. ACV, TCV and billings all differ.
  • New vs installed base. ASP describes new deals; ARR/customer includes years of accumulated expansion and is therefore always higher in a healthy business.
  • Ramped deals. Contracts that step up over the term have no single ACV.
  • Consumption pricing. Committed vs actual consumption produces two defensible ACVs for the same contract.

What it does to the number

This matters more than it looks, because ACV is the primary segmentation axis in this entire corpus. Choosing ARR/customer instead of new-deal ACV can move you a full band up the table and hand you the wrong benchmark for everything else.

Source conventionsKeyBanc reported a $62K median ACV (2024 survey). The Bridge Group reported a $47K median ASP (2024) and a $50K median ASP in its 2025 SDR panel. SaaS Capital and High Alpha both band by ACV as respondents self-report it.

"Is that new-deal ACV, or ARR divided by customers?"

Retention & churn metrics

Net revenue retention

Highest mismatch risk NRR · NDR · net dollar retention · net expansion rate

Canonical — SaaS Capital's published formula, the clearest in circulation MRR in Dec Y2 from customers who were customers in Dec Y1 ÷ total MRR in Dec Y1
Includes upsell, cross-sell, price increases, downgrades and churn. Can exceed 100%.

Variants that change the number

  • New customers in the denominator. They must be excluded. Including customers acquired during the period is the most common outright error and inflates the result.
  • Churned logos dropped. Churned customers must stay in at $0. Dropping them turns NRR into an expansion rate and can add 10+ points.
  • Annual cohort vs monthly annualised. Compounding a monthly rate produces a different number than an annual point-to-point cohort on identical data.
  • ARR vs MRR vs recognised revenue basis.
  • Trailing-twelve-month vs point-in-time snapshot.
  • Public-company convention. "Dollar-based net expansion rate" in 10-Ks often uses a trailing-12-month cohort and is not directly comparable to private survey NRR.

What it does to the number

The two failure modes both push the same way: up. A company reporting 115% NRR with new customers in the denominator and churned logos dropped may be running 95% on the canonical formula. When an NRR figure looks anomalously strong for its ACV band, check these two things first.

Source conventionsSaaS Capital publishes the formula above explicitly. High Alpha defines NRR as "annual net revenue retention (after churn, inclusive of upsells and expansion) seen in cohorts". ICONIQ reports NDR from portfolio operating data rather than survey self-report, which is why its figures run higher and are published as quartiles.

"Are churned customers still in the numerator at zero, and are new customers out of the denominator?"

Gross revenue retention

Low mismatch risk GRR · gross dollar retention

Canonical Same cohort formula as NRR, with each customer's ending value capped at its starting value. Captures churn and downgrade only. Cannot exceed 100%.

Variants that change the number

  • Capping per customer vs in aggregate. Capping only the total lets one customer's expansion mask another's downgrade — this is the error that produces a "GRR" above 100%.
  • Downgrades excluded. Some companies count only full churn, which makes GRR a logo metric wearing a revenue metric's name.
  • Renewal-base GRR. Measured only against contracts that actually came up for renewal in the period, rather than the whole book. Materially higher for multi-year books.

What it does to the number

Renewal-base GRR is the flattering variant and is common in board decks at companies with multi-year contracts, because contracts that never came up for renewal cannot churn. A GRR above 100% is definitionally impossible — if you see one, the cap is being applied in aggregate.

Source conventionsSaaS Capital and High Alpha both use the whole-book cohort convention. Both land on ~90–92% as the all-company median, which is the strongest agreement between any two panels in this corpus.

"Is that against the whole book, or only against what came up for renewal?"

Churn rate

Highest mismatch risk

Two different metrics with one name Logo churn = customers lost ÷ customers at period start  ·  Revenue churn = revenue lost ÷ revenue at period start

Variants that change the number

  • Logo vs revenue. In an SMB-weighted book, logo churn can be double revenue churn, because the customers who leave are the small ones.
  • Monthly vs annual, converted wrong. 2% monthly churn is not 24% annually. It is 1 − 0.9812 = 21.5%. Linear annualisation overstates.
  • Denominator timing. Start-of-period, average, or end-of-period base.
  • Is churn the complement of GRR? Only if downgrades are treated identically in both.

What it does to the number

The monthly-to-annual conversion error is small at low churn and large at high churn: at 5% monthly, linear gives 60% and the correct compounding answer is 46%. Any SMB benchmark quoted as an annual figure should be checked for which way it was derived.

Source conventions — the sources in this corpus report revenue retention rather than churn, and on an annual cohort basis. SaaS Capital states plainly that it considers dollar-based retention the more important metric and surveys on that basis.

"Logos or dollars? And was the annual number compounded or multiplied?"

Efficiency metrics

CAC payback period

Highest mismatch risk

Canonical — gross-margin-adjusted fully-loaded S&M cost to acquire a customer ÷ (new MRR × subscription gross margin)  — expressed in months.

Variants that change the number

  • Gross margin omitted. Dividing by raw revenue rather than gross profit understates payback by roughly the gross-margin haircut — about 20–25% at typical SaaS margins.
  • New-only vs blended. Including expansion ARR in the denominator shortens payback substantially and measures a different thing.
  • Cost scope. Sales only, S&M, or S&M plus onboarding and customer success.
  • Unallocated founder and support cost. High Alpha flags this explicitly as an early-stage distortion.
  • Period lag. Prior-period spend against current-period ARR vs same-period both.

What it does to the number

Every common variant pushes payback down. The sub-$1M ARR median of 5 months in the High Alpha data is not a real 5 months — the publisher itself warns that early-stage companies under-allocate founder salaries, support costs and onboarding into CAC.

Source conventionsHigh Alpha defines it as "months of subscription gross margin to recover the fully-loaded cost of acquiring a customer" — the strict version — and publishes quartiles by ARR band.

"Is that gross-margin-adjusted, and does the denominator include expansion?"

CAC ratio

Medium mismatch risk

Three ratios, and they are not interchangeable New CAC ratio = total S&M ÷ new customer ARR
Blended CAC ratio = total S&M ÷ (new customer ARR + expansion ARR)
Expansion CAC ratio = expansion-attributed S&M ÷ expansion ARR

Why it is worth using

  • It converts directly into a budget: $25M of new ARR at a 1.5 New CAC ratio requires a $37.5M go-to-market budget. Payback period does not do this.
  • It is denominated in dollars of ARR, not logos, so it is stable across mixed deal sizes.
  • Blended will always look better than New. A company quoting an unqualified "CAC ratio" is usually quoting Blended.

Published benchmark

Benchmarkit put the median New CAC ratio at $2.00 — two dollars of sales and marketing per dollar of new customer ARR — up 14% year over year, with the fourth quartile at $2.82.

Source conventionsBenchmarkit is the source that maintains this family of definitions most rigorously and segments them by ARR, ACV, GTM motion and VC- vs PE-backed.

"New or blended?"

Rule of 40

Medium mismatch risk

Canonical YoY growth rate (%) + profitability margin (%)  ≥ 40

Variants that change the number

  • Which margin. Free cash flow margin, EBITDA margin, or operating margin. FCF and EBITDA can differ by many points at companies with heavy deferred revenue or capitalised development.
  • ARR growth vs recognised revenue growth. ARR growth leads revenue growth, so the ARR version reads higher during acceleration and lower during deceleration.
  • Trailing twelve months vs annualised current quarter.

What it does to the number

Companies choose the flattering combination, and it is rarely disclosed. Treat any single-company Rule of 40 claim as unverifiable unless both terms are stated.

Source conventionsHigh Alpha defines it as "year-over-year ARR growth percentage plus last twelve months free cash flow margin or EBITDA margin", explicitly permitting either margin.

"Growth of what, plus which margin?"

S&M spend and R&D spend

Medium mismatch risk

Canonical function spend including headcount ÷ ending ARR for the period.

Variants that change the number

  • Denominator: ending ARR vs recognised revenue vs average ARR. At 30% growth, ending-ARR and average-ARR denominators differ by roughly 13%.
  • Is customer success inside S&M? Placement varies and moves several points.
  • Is RevOps inside S&M, G&A, or split?
  • Fully-loaded vs cash comp only.

What it does to the number

Function-boundary choices dominate. Before comparing an S&M ratio, confirm whether CS and RevOps sit inside it — that alone can account for the entire apparent gap.

Source conventionsHigh Alpha uses spend including headcount as a percentage of ending ARR. LeanScale's own RevOps investment study works through the same denominator problem for RevOps specifically and reaches the same conclusion: pick the denominator before the number.

"Percent of ending ARR or of revenue, and is CS inside the numerator?"

Magic number

Medium mismatch risk

Canonical (quarterly) (current-quarter ARR − prior-quarter ARR) × 4 ÷ prior-quarter S&M spend

Variants that change the number

  • Net new ARR vs new-customer ARR only. The net version includes expansion and churn and is the standard.
  • Same-quarter vs prior-quarter S&M. Prior-quarter is correct — spend leads revenue — but same-quarter is common.
  • Gross-margin adjusted or not.

What it does to the number

Magic number and CAC ratio are reciprocals of each other under matching conventions — a New CAC ratio of $2.00 is a magic number of 0.5. If a company's two figures do not reconcile, the conventions differ somewhere.

"Prior-quarter spend, and net new or new-customer ARR?"

ARR per employee

Low mismatch risk

Canonical ending ARR ÷ full-time employees at period end

Variants that change the number

  • Contractors and offshore teams. Excluding them is the standard distortion, and it is growing as offshore delivery scales.
  • FTE headcount vs FTE-equivalent.
  • Period-end vs average headcount during a hiring year.

What it does to the number

Cleanest metric in the corpus definitionally, and increasingly the one people most want, because it is the crispest read on whether AI leverage is real.

Source conventionsHigh Alpha counts full-time employees at quarter end and publishes medians and upper quartiles by ARR band.

"Are contractors and offshore staff in that headcount?"

Software gross margin

Low mismatch risk

Canonical (subscription revenue − subscription COGS) ÷ subscription revenue

Variants that change the number

  • Subscription-only vs total (blended with services). Services drag blended margin down sharply; the subscription-only figure is the benchmark standard.
  • What sits in COGS. Hosting, support, customer success, professional services and — newly material — AI inference cost.

What it does to the number

Inference cost is the live issue. High Alpha's 2025 data shows early-stage gross margins declining year over year while mature-company margins held, consistent with AI-native products carrying real variable cost of goods.

"Subscription-only or blended with services, and is inference in COGS?"

Funnel benchmarks

Conversion, cycle time and rep productivity.

The strongest by-ACV funnel data available comes from ICONIQ's annual GTM survey. Read every figure here against the definition cards above — particularly win rate and stage conversion, where the published averages blend measurement conventions.

Sales cycle by ACV

Average weeks · ICONIQ Growth, State of Go-to-Market 2026 (Jan 2026 survey, 157 respondents on this chart)

< $10K
6 wks
$10K–$50K
12 wks
$50K–$100K
17 wks
$100K+
24 wks

Per-band n: 9 · 30 · 35 · 83. The <$10K cell is thin and should be read as directional only. Shape: the cycle roughly doubles for each ~5× step in ACV. Overall average cycle fell from 25 weeks (H1 2025) to 19 weeks (H2 2025), against 23 weeks in 2024 — but over the same window the share of new-logo contracts under one year rose from 2% to 13%. Customers are signing faster for shorter terms.

Stage conversion rates

Average · ICONIQ Growth, State of Go-to-Market 2026 · n=147 (2026), n=167 (2025)

Stage transition20262025Change
New lead → MQL28%25%+3
MQL → SQL30%27%+3
SQL → closed-won28%25%+3
Demo → closed-won38%36%+2
Free trial / POC → paid50%36%+14
The POC jump is the largest single-year move in the dataset. ICONIQ attributes it to POCs being run as a disciplined motion rather than an open trial — support level scales with ACV, with a dedicated solutions architect on 49% of POCs above $100K ACV versus 25% below $50K.

Win rate by opportunity source

Average · closed-won ÷ (closed-won + closed-lost) · ICONIQ Growth · n=145 (2026), n=165 (2025)

Opportunity source20262025
Customer success52%
Sales43%38%
Channel / partner39%35%
Marketing27%23%
This denominator excludes no-decision outcomes entirely — see the win-rate definition card. Use these figures for the relative ranking by source, which is robust, rather than as an absolute target.
Pipeline sourcing High-growth companies in the ICONIQ panel derive 60–80% of pipeline from sales and channel motions against 15–20% from marketing — a materially more seller-led mix than their non-high-growth peers. "High growth" is defined by ICONIQ as 200%+ YoY at $10–25M revenue, 100%+ at $25–100M, 50%+ at $100–250M, and 30%+ at $250M+.

Quota attainment

Share of reps achieving quota · two sources, two bases

BasisFigureSourceNote
Ramped AEs, 202562%ICONIQUp from 58% (2024) and 59% (2023)
SMB AEs68%ICONIQn=71
Mid-market AEs59%ICONIQn=113
Enterprise AEs64%ICONIQn=125
Strategic AEs61%ICONIQn=106
All AEs, 202648%Bridge GroupDown from 51% (2024), 66% (2022)
All AEs, high AI engagement57%Bridge Groupvs 39% in the lowest tercile

Rep-level economics

The Bridge Group — 2026 AE Research (n=158) and 2025 SDR Research (n=351; 83% B2B SaaS, $47M median revenue, $50K median ASP)

MetricCurrentPriorDirection
AE median quota$960K$800K (2024)Quota CAGR ~2.4%
AE median OTE$200K$190K (2024) · $167K (2022)OTE CAGR ~4.9%
Quota-to-OTE ratio4.6×4.2× (2024)Rising
AE ramp time6.2 mo5.7 mo (2024)Highest in study history
Experience required at hire3.7 yrs2.7 yrs (2022)Rising sharply
AE win rate19%23% (2022)2024 figure; broader denominator
SDR : AE ratio1 : 2.41 : 2.4Flat since 2018
SDR ramp3.0 moLowest since 2010
SDR tenure1.9 yrsHighest since early 2010s
SDR monthly quota (meetings held)10Down ~40% since 2018
SDR daily activities11244 phone · 41 email · 19 LinkedIn · 8 other
SDR quality conversations / day4.1First rebound in study history
Pipeline generated per SDR$3.78M$2.83M (2022)Rising
SDR median OTE$80K$80K (2022)68:32 base:variable
SDR annual attrition40%13% involuntary · 11% voluntary · 16% promotion
ACV is the largest single determinant of AE quota in the Bridge Group data: the gap between a sub-$25K-ACV seller and a $250K+ seller is roughly 2.5×.
Retention benchmarks

NRR and GRR, banded by ACV.

Retention is the best-instrumented area in this corpus: two independent panels of 800–1,500 companies, both publishing by ACV, both stating their formulas. Where they agree, confidence is high. Where they disagree — the top of the ACV range — the disagreement is itself informative.

Net and gross revenue retention by ACV

Median · SaaS Capital, 2023 B2B SaaS Retention Benchmarks (12th annual survey, 1,500+ private B2B SaaS companies, excludes <$1M ARR)

ACV bandNRR — 25thNRR — medianNRR — 75thGRR — median
Less than $12K93%100%105%90%
$12K – $25K96%102%108%90%
$25K – $50K97%103%111%92%
$50K – $100K98%105%113%93%
$100K – $250K101%107%118%93%
More than $250K103%110%120%93%
All-company medians: NRR 102%, GRR 91%. SaaS Capital's own guidance: below $25K ACV, 90% GRR is the norm; above it, benchmark to 93%. GRR is treated as table stakes — below 90% and growth falls under the population median.

Why ACV is the right cut. In SaaS Capital's words, companies that share a similar selling price have the most in common — more than company age, revenue level or industry. They organise similarly, go to market similarly, and support customers similarly.

The same cut, a different panel

Median · High Alpha, 2025 SaaS Benchmarks Report (9th annual, 800+ respondents)

ACV band% of panelYoY growthGRRNRR
Less than $1K4%30%83%98%
$1K – $5K11%25%88%97%
$5K – $10K8%25%85%94%
$10K – $15K10%40%88%107%
$15K – $25K11%38%88%104%
$25K – $50K16%30%92%105%
$50K – $100K18%44%94%104%
$100K – $250K12%30%91%102%
More than $250K9%32%92%105%
High Alpha's reading: performance converges in the mid-market. Companies between $10K and $100K ACV show the strongest blend — growth above 30%, GRR near or above 90%, NRR above 104%. Sub-$10K contracts churn and cannot expand; at the top end growth slows as cycles lengthen and concentration rises.

Cross-check

The two panels agree closely on GRR — roughly 90% below $25K ACV rising to 92–94% above it — which is the strongest corroboration in this corpus. They disagree on the shape of NRR at the top end: SaaS Capital finds NRR rising monotonically to 110% above $250K, while High Alpha finds it peaking in the $10K–$100K range and flattening above. Treat the direction (higher ACV, higher retention) as well-established and the exact top-end value as unsettled.

Retention by ARR band

Median and quartiles · High Alpha 2025 (n=800+)

ARR bandNRR — lowerNRR — medianNRR — upperGRR — median
Less than $1M78%100%116%92%
$1M – $5M91%104%110%92%
$5M – $20M95%103%115%88%
$20M – $50M98%103%110%90%
Greater than $50M101%101%108%88%
Note the spread, not just the median: sub-$1M companies range from 78% to 116%, and the range tightens steadily with scale. Early-stage NRR is a weak signal.

Retention against growth, contract length and funding

Median · SaaS Capital (n=1,500+)

CutGroupNRRGRR / other
Growth rate by NRRNRR below 90%20% median growth
NRR 90–100%26%
NRR 100–110%35%
NRR 110–120%40%
NRR 120–130%49%
NRR above 130%70%
Contract lengthMonth-to-month100.0%89.5%
Annual101.0%90.0%
Multi-year105.5%95.0%
FundingBootstrapped100%91%
Equity-backed103%91%
Product shapeHorizontal102%90%
Vertical101%92%
Company ageUnder 3 years102%92%
3–5 years103%94%
6–14 years101–102%90%
Population median growth in that survey was 34%. Two cautions: young-company GRR is artificially elevated because customers have not yet had a chance to churn — the decline from 94% to 90% between years 3–5 and 6+ is a measurement artefact maturing, not performance decaying. And the multi-year retention premium may carry the same artefact: contracts that have not come up for renewal cannot churn.
Most recent readings SaaS Capital 2026 (bootstrapped, $3–20M ARR, 1,000+ companies): median growth 15% — down from 20% — with 42.3% at the 90th percentile; NRR 103% median, 117.9% at p90; GRR 91% median, 100% at p90. KeyBanc / Sapphire, 16th annual (Nov 2025): ARR growth expected to accelerate from 15% in 2024 to 20% in 2025; gross retention recovering toward 90% from 86% in 2023; net retention above 100%. ICONIQ: top-quartile net dollar retention in the ~110–120% range across revenue bands, from portfolio operating data rather than survey self-report.
Efficiency benchmarks

What it costs to acquire, and what the business keeps.

Full operating benchmarks by ARR band

Median [lower quartile – upper quartile] · High Alpha, 2025 SaaS Benchmarks Report (n=800+)

Metric<$1M ARR$1–5M$5–20M$20–50M>$50M
YoY growth rate100% [29–300]50% [24–100]31% [15–72]30% [16–41]16% [10–29]
CAC payback (months)5 [2–8]8 [5–14]14 [8–22]20 [11–27]17 [13–22]
Gross revenue retention92% [80–100]92% [83–95]88% [82–95]90% [85–95]88% [84–90]
Net revenue retention100% [78–116]104% [91–110]103% [95–115]103% [98–110]101% [97–108]
S&M spend (% of ARR)28% [20–50]30% [20–50]30% [22–53]29% [24–34]25% [20–40]
R&D spend (% of ARR)43% [23–76]38% [24–50]31% [20–54]32% [23–40]30% [25–40]
Software gross margin74% [60–80]77% [60–85]80% [70–86]78% [71–84]79% [70–83]
Rule of 4046% [25–98]33% [10–80]20% [−3–35]24% [11–41]30% [15–38]
Employees (median)9 [5–12]22 [15–35]66 [48–92]131 [100–223]361 [263–729]
ARR per employee — median$55,556$136,364$166,667$268,235$277,778
ARR per employee — upper quartile$100,000$200,000$220,588$350,000$396,635
Panel: ARR <$1M 23%, $1–5M 27%, $5–20M 31%, $20–50M 10%, >$50M 9%. 69% US. ICP: enterprise 37%, mid-market 35%, SMB 19%. 50% of respondents were CEO or founder. CAC payback caution from the publisher itself: early-stage companies frequently fail to allocate founder salaries, support costs and onboarding into CAC, artificially lowering the reported payback period.
CAC ratio Benchmarkit's median New CAC ratio is $2.00 — two dollars of sales and marketing spend per dollar of new-customer ARR — up 14% year over year, with the fourth quartile at $2.82. This is the most directly plannable efficiency metric in the corpus: it converts straight into a budget. $25M of new ARR at a 1.5 ratio requires a $37.5M go-to-market budget.
RevOps investment LeanScale's own reconciliation of eleven published sources puts RevOps at 5–10% of go-to-market headcount (median 7%), or roughly one RevOps head per $21M of ARR — and about 0.9% of revenue, not the 5–10% of revenue the rule of thumb is usually misquoted as. Full working at leanscale-revops-investment.netlify.app.
Reconciliation

Five places credible sources disagree, and why.

Each of these is a real conflict between reputable panels. In every case the disagreement resolves to measurement rather than to one source being wrong — which is exactly why the definition layer exists.

Win rate: 43% or 19%?

43%ICONIQ · sales-sourced · 2026
19%Bridge Group · median · 2024

ICONIQ's published formula excludes every deal that did not reach a won/lost decision. In enterprise sales, no-decision is routinely the largest single loss category, so removing it lifts the rate substantially. The Bridge Group measures on a broader opportunity base closer to what a CRO experiences on a forecast call.

Panel composition compounds it: ICONIQ's respondents skew to $100K+ ACV growth-stage companies with mature qualification, and 42% of the panel sits in the $100K–$500K ACV band.

Use: ICONIQ for relative comparisons between opportunity sources. Bridge Group for an absolute expectation of what a rep converts. Never compare your own number to either without first establishing which denominator you compute on.

Quota attainment: 62% or 48%?

62%ICONIQ · ramped AEs · 2025
48%Bridge Group · all AEs · 2026

ICONIQ measures ramped AEs only. At a company hiring into growth, a meaningful share of the rep base is ramping at any moment and structurally cannot attain — excluding them raises the figure. The two also cover different years, and the trend is downward.

Use: the ramped-only figure when assessing whether your tenured reps are performing; the all-rep figure when modelling capacity and forecast coverage.

NRR: 102% or 120%?

102%SaaS Capital · median · all
103%High Alpha · median · $5–20M
110–120%ICONIQ · top quartile

This is not a disagreement at all — it is the most common benchmarking error in circulation. SaaS Capital and High Alpha publish medians across their whole panel. ICONIQ publishes top-quartile net dollar retention from portfolio operating data. Comparing your median against a published top quartile will always make you look broken.

SaaS Capital's own top quartile reaches 118–120% above $100K ACV — which reconciles cleanly with the ICONIQ figure once you compare like with like.

Use: match the statistic first. If a source does not say whether a figure is a median or a quartile, do not use it.

Does NRR keep rising above $100K ACV?

YesSaaS Capital · 110% above $250K
NoHigh Alpha · peaks $10–100K

A genuine, unresolved disagreement between two large panels on the same cut. SaaS Capital finds NRR rising monotonically with ACV; High Alpha finds it peaking in the $10K–$100K range and flattening or dipping above. Both agree GRR rises with ACV.

Plausible reconciliations: different survey years and macro conditions; High Alpha's panel skews earlier-stage, so its high-ACV cell may capture young enterprise sellers rather than mature ones; and per-band sample sizes at the top end are small in both.

Use: treat "higher ACV, higher retention" as well-established through $100K. Above $100K, hold the estimate loosely and weight your own data more heavily.

Are sales cycles getting shorter?

−6 wksICONIQ · 25→19 wks, H1→H2 2025
LongerBridge Group · majority report increases

ICONIQ's average cycle fell six weeks across 2025. The Bridge Group's 2026 AE research reports that near-majorities saw increases in stakeholder count, cycle length, discounting pressure and deal slippage against Q1 2025.

The likely reconciliation sits in ICONIQ's own data: over the same period the share of new-logo contracts under one year rose from 2% to 13%, and three-year contracts fell from 36% to 23%. Shorter, smaller, more reversible commitments close faster. That is a change in what is being bought, not necessarily an improvement in how it is sold.

Use: never read a cycle-time improvement without checking contract length and deal size alongside it. Cycle time falling while contract duration falls is a mix shift.
Source directory

Who publishes what, and what each one is good for.

Every figure in this corpus traces to one of these. All seven are free at time of writing. Cite the publisher, the edition year and the sample — never this page alone.

ICONIQ Growth

The State of Go-to-Market · annual

Edition
March 2026, from a January 2026 survey
Sample
150+ B2B software GTM executives (CRO, Head of Sales, CEO, Head of RevOps)
Panel
$10M–$500M+ revenue. ACV: 42% at $100K–$500K, 22% at $50–100K, 20% at $10–50K. 62% sales-led, 31% hybrid, 7% PLG
Segments by
ACV, revenue band, growth rate, motion

Best for: sales cycle and funnel conversion by ACV. The only source publishing a clean by-ACV cycle-time curve.

Bias: portfolio-adjacent and upmarket. Flatters anyone below $25K ACV. Per-chart n falls as low as 9 in the lowest ACV band. Publishes averages, not medians, for funnel metrics.

iconiq.com/growth/reports/state-of-go-to-market-2026

SaaS Capital

Annual survey + retention research briefs

Edition
Retention Brief 28 (12th annual); 2026 spending benchmarks
Sample
1,000–1,500+ private B2B SaaS companies, largest of its kind
Panel
Private B2B SaaS incl. bootstrapped; excludes <$1M ARR from most cuts
Segments by
ACV, ARR, age, funding type, contract length, vertical/horizontal

Best for: retention by ACV. The only source with enough n to publish NRR quartiles by ACV band, and it states its formula explicitly.

Bias: a growth-debt lender, so the respondent pool skews toward companies with real revenue and reasonable retention. Self-reported.

saas-capital.com/research

High Alpha

SaaS Benchmarks Report · annual · 9th edition

Edition
2025 (published late 2025)
Sample
800+ respondents; ~5,000 companies cumulatively over nine years
Panel
23% <$1M ARR, 27% $1–5M, 31% $5–20M, 10% $20–50M, 9% >$50M. 69% US. 50% CEO/founder
Segments by
ARR, ACV, GTM channel, pricing model

Best for: the full operating picture at the early and growth stages, with quartiles on every metric and an explicit definitions page.

Bias: earlier-stage and PLG-weighted; the natural complement to ICONIQ's upmarket panel. This is the continuation of the OpenView benchmarks — Kyle Poyar took the report to High Alpha after OpenView wound down.

highalpha.com/saas-benchmarks

The Bridge Group

AE and SDR Metrics research · biennial · since 2007

Edition
2026 AE research (10th); 2025 SDR research (10th)
Sample
158 companies (AE); 351 companies (SDR)
Panel
83% B2B SaaS, 78% North America, $47M median revenue, $50K median ASP
Segments by
ACV band, segment, motion

Best for: rep-level economics. Quota, ramp, attainment, OTE, SDR:AE ratio, activity levels — nothing else covers this ground, and the time series runs to 2007.

Bias: the most conservative panel here; its win rate and attainment figures run well below the VC-published surveys. Arguably the most representative of an ordinary company.

bridgegroupinc.com/research

Benchmarkit

SaaS Performance Metrics + B2B Marketing Benchmarks · annual

Edition
2025 performance metrics; 2026 marketing benchmarks
Sample
Not consistently disclosed per chart
Segments by
ARR, ACV, GTM motion, target customer, product category, VC- vs PE-backed

Best for: definitional rigour on efficiency metrics. Maintains the new/blended/expansion CAC ratio family properly and has an interactive filter by company profile.

Bias: per-chart sample sizes are the least transparent of the seven. Strong on finance metrics, thinner on funnel.

benchmarkit.ai

KeyBanc Capital Markets + Sapphire Ventures

Private Company SaaS Survey · annual · 16th edition

Edition
November 2025
Sample
Several hundred private SaaS companies; median ACV $62K in the 2024 edition
Segments by
ACV, growth rate, ARR, GTM motion

Best for: the P&L context around the funnel — CAC ratio, magic number, S&M as a share of revenue, retention, and the longest-running private SaaS operating time series.

Bias: an investment bank's survey; respondents skew toward companies of banking interest. Headline findings are public; the full detail requires requesting the report.

key.com · SaaS survey

LeanScale

Field studies from a delivery panel of 40–65 B2B software companies

Editions
GTM Tech Stack (50+ cos), AI Workflows in GTM (40+ cos), RevOps Investment (11 reconciled sources)
Method
Provenance-verified sweeps of delivery records, customer calls and shared Slack channels — observed behaviour rather than self-report

Best for: what companies actually run, as opposed to what they report on a survey. The only non-self-reported panel in this corpus.

Bias: LeanScale's own customer base — growth-stage B2B software that has engaged a RevOps partner. Smaller n than the survey houses.

knowledge.leanscale.team
Methodology

What this is, and how to cite it.

01

A reconciliation, not a survey.

No LeanScale customer data underlies the benchmark figures on this page. Every number is attributed to a named third-party publisher, edition and sample. Where LeanScale's own measurement appears it is labelled as such.

02

Primary documents only.

Figures were taken from the publishers' own PDFs, research briefs and press releases. Search results for "SaaS benchmarks" are now dominated by AI-generated aggregator sites that invent precise-sounding numbers with no survey behind them; none were used, and none should be.

03

Every source here is self-reported except one.

Survey panels are opt-in, and companies having a bad year do not fill out the survey. Assume a consistent upward bias across all of it. Billing-data sources read low for the opposite reason; bracketing the two is better than trusting either.

04

Definitions are stated where the publisher stated them.

SaaS Capital, High Alpha and ICONIQ publish formulas; those are quoted. Where a publisher does not define a metric, the definition card says so and the figure is flagged as convention-blended rather than presented as precise.

05

Sample size is reported at the cell, not the report.

Where a publisher discloses per-chart n, it is carried through — including where it is uncomfortably small, such as the nine companies behind the sub-$10K ACV sales-cycle figure.

06

How to cite.

Cite the original publisher, edition year and sample — for example: "ICONIQ Growth, State of Go-to-Market 2026 (January 2026 survey, n=157)". This corpus is a finding aid and a definition layer, not a primary source.

The standing recommendation

Use this corpus to answer "are we structurally normal for our price point?" For "what should we fix?", compute your own stage conversion on an entered-cohort basis and compare the company against its own trailing quarters. That comparison is the only one where the definition is guaranteed constant — which is the whole argument of this page.