B2B SaaS Funnel Conversion Benchmarks by Stage (2026)
Somewhere between your website visitor and your closed deal, roughly 99.7% of the funnel disappears. That is not failure - it is the benchmark. The useful question is where yours disappears faster than it should.
This is the hub post for our benchmarks series: every stage of the B2B SaaS funnel - visitor to lead, lead to MQL, MQL to SQL, SQL to opportunity, opportunity to close - from three labeled datasets, the conflicts between them shown rather than smoothed, and the math for finding your weakest stage. Each stage links to our deeper dive on that number.
The stage-by-stage table
Three sources, side by side. Labels matter: all three are agency data with different mixes and definitions.
| Stage | Powered by Search ($10M-$100M ARR SaaS) | GrowthSpree (B2B SaaS composite) | First Page Sage (30 industries, 65% B2B, 2017-2025) |
|---|---|---|---|
| Visitor -> Lead | 1.4% | 2-5% (top 10%: 8-15%) | - |
| Lead -> MQL | 41% | 40-60% | 17-45% |
| MQL -> SQL | 39% | 25-40% | 32-58% |
| SQL -> Opportunity | 42% | 50-70% | 40-66% |
| Opportunity -> Close | 39% | 15-25% | 37-66% (SQL->closed) |
| Full funnel (visitor -> customer) | 0.29% | 0.1-0.5% (top: 1-2%) | - |
At enterprise scale the funnel compresses further: Powered by Search measures 0.7% visitor-to-lead and a 0.10% full funnel at $1B+ ARR companies.
Look at the Opportunity-to-Close column: 39% versus 15-25% versus 37-66%. These are credible sources describing the “same” stage with barely overlapping ranges - and GrowthSpree’s funnel report even quotes a different MQL-to-SQL band (25-40%) than its own dedicated MQL study (18-22% average, which we cover in depth in the MQL to SQL benchmarks). The lesson is not that benchmarks are useless; it is that stage definitions differ everywhere. Benchmark against ranges, and against your own trailing quarters above all.
The full-funnel math, worked
Take 10,000 monthly visitors through the mid-band rates: ~2% become leads (200), ~45% of those become MQLs (90), ~30% become SQLs (27), ~50% become opportunities (13-14), ~25% close. Three to four customers from ten thousand visitors - squarely inside the 0.1-0.5% benchmark band.
Now the part that makes this table worth a bookmark: those stages multiply. Improve any single stage 30% and the whole funnel improves 30%. Improve two stages 30% and it compounds to 69% more customers from the same traffic - which is why the right question is never “how do we get more traffic” until you know which stage is leaking. (Traffic is also getting more expensive: see the Google Ads and LinkedIn Ads benchmarks.)
Channel changes everything
The single most striking split in Powered by Search’s data is by channel, full-funnel:
| Channel | Visitor -> Lead | MQL -> SQL | Full funnel |
|---|---|---|---|
| SEO / organic | 2.1% | 51% | 1.90% |
| PPC | 0.7% | 26% | 0.27% |
A visitor who arrived by searching converts to a customer at 7x the rate of a paid-click visitor, and the gap widens at every stage, not just the first. That does not make paid wrong - it makes blended funnel reporting misleading. Track your funnel per channel, or your “average” will hide one channel subsidizing another. This mirrors what the MQL-level data shows about intent-born versus interruption-born leads, and it is the quantified case for owning your demand capture before scaling interruption.
Fixing your weakest stage: the map
Once your per-stage rates are on the table, work the stage furthest below band - cheapest fixes first:
- Visitor -> Lead below ~1.5%? Your pages ask for contact info without answering questions. Start with the website conversion benchmarks and demo request benchmarks; the fastest fix is usually chat that answers the blocking question instead of a fourth form field. GrowthSpree cites Unbounce 2026 data showing custom-built landing pages converting at 11.6% versus 3.8% for templates - intent-matching is the lever.
- Lead -> MQL below ~35%? You are generating contacts, not prospects - revisit which offers earn contact info and score on behavior.
- MQL -> SQL below ~20%? Definitions and follow-up speed, in that order. The full MQL-to-SQL breakdown covers both; answering within minutes is the cheapest lever and the one most teams lose by default. If demos are booked but not happening, that is its own leak: demo no-show benchmarks.
- SQL -> Opportunity or Close below band? This is sales territory, but marketing owns two inputs: the quality of what got stamped SQL, and the pipeline coverage math that determines whether reps are working thin.
- Everything slightly below band? Check the money math instead: CAC payback and marketing-sourced pipeline tell you whether the funnel is economically fine despite unglamorous rates.
One adjacent benchmark for PLG motions, via Powered by Search’s roundup: free trials convert to paid at 8-12% on average, freemium at 3-7% (referencing Totango and Lenny’s Newsletter survey data). If that is your motion, the free trial email sequence is the stage-fixer.
How to use these numbers without fooling yourself
- Define stages in writing first. The source conflicts above exist because “MQL” and “opportunity” mean different things in different CRMs. Yours needs one definition, agreed with sales, before any benchmark means anything.
- Segment by channel and motion. A blended funnel mixing SEO with paid social, or PLG with sales-led, averages two different businesses into one meaningless number.
- Watch trends inside bands. Being at 25% MQL-to-SQL matters less than being at 25% and falling.
- Measure from traced data, not vibes. Most teams cannot actually compute this table for themselves because touchpoints live in five tools. Wiring revenue attribution to the exact message - and tracking marketing-sourced pipeline without a RevOps team - is what turns this post from reading into a dashboard.
Frequently asked questions
What percentage of visitors become customers?
Benchmarks cluster at 0.1-0.5% full-funnel, with 0.29% measured for $10M-$100M ARR SaaS and 1-2% for top performers. One in ~300 is normal.
What is a good visitor-to-lead rate?
1.4% measured by Powered by Search; 2-5% typical and 8-15% top-decile per GrowthSpree. The spread is definitional - decide what counts as a lead first.
Which stage should I fix first?
The one furthest below its band, cheapest first. The edges - page conversion and follow-up speed - are usually marketing’s fastest wins because they compound through every stage after them.
Why do sources disagree?
Different account mixes, definitions, and denominators - even within one agency’s two reports. Use ranges to find your outlier stage; use your own trailing data as the real baseline.
The bottom line
The benchmark funnel loses 99.5%+ of visitors, top performers lose “only” 98%, and the difference between those two numbers is the entire growth story of a SaaS company. You cannot out-spend a leaky funnel at 2026 traffic prices - but because stages multiply, you rarely need to fix more than the one or two where you are genuinely below band. Find them with per-channel, per-stage data; fix the edges first; and let the compounding do what more budget cannot.
See the edge-stages live: Marqeable turns visitors into leads with AI website chat that answers instead of gating, follows up on every new lead in minutes with automations, and computes your real funnel - dollars tied to the exact message - with attribution.
Marqeable runs your campaigns, answers every visitor, text, and email in seconds, and turns them into booked jobs and meetings - even at 9pm on a Saturday. We’re in private beta with a small early cohort. Get early access
