# The SaaS marketing funnel

> Every funnel stage with the conversion rate that counts as normal, why published SaaS website conversion spans 1.1% to 7.6%, and which leak to fix first.

Source: https://saas-marketing.net/guides/saas-marketing-funnel/
Topic: SaaS Marketing
Type: guide
Published: 2026-09-11
Last updated: 2026-09-11
Publisher: SaaS Marketing (saas-marketing.net)
License: CC BY 4.0. Quote or republish with attribution and a link to https://saas-marketing.net/guides/saas-marketing-funnel/

## Short answer

A SaaS marketing funnel has seven stages: visitor, lead, MQL, SQL, opportunity, customer and expansion. Typical rates run 1% to 5% visitor to lead, 25% to 45% MQL to SQL, and 15% to 30% opportunity to closed won. Trial to paid splits by design: opt-in trials convert near 9% while opt-out trials that take a card up front convert near 44%. Publish the event definition beside every number or the benchmark means nothing.

## Key takeaways

- Published SaaS website conversion rates span 1.1% to 7.6% because each source counts a different event as a conversion.
- Opt-in trials convert to paid near 9% and opt-out trials near 44%, so the trial design matters more than the funnel copy.
- Diagnose leaks from the bottom of the funnel upward, because fixing the top multiplies whatever waste already exists.
- Landing 10 new customers at typical mid market rates needs roughly 860 leads and about 34,000 qualified sessions.
- One named event per stage, with the definition published next to the number, is the only way a benchmark stays usable.
- The bowtie extension past closed won catches the business that reports a good quarter while losing net revenue.

---

Two SaaS companies can report the same funnel stages, the same stage names and wildly different conversion rates, and both can be telling the truth. The reason is boring and fixable: they are counting different events. Until you write down which single event marks each stage, a benchmark is decoration. This page gives the stages, the ranges that count as normal, the definition that has to sit beside each one, and the order to fix leaks in.

## The seven stages, and the one event that defines each

A B2B SaaS funnel runs visitor, lead, marketing qualified lead, sales qualified lead, opportunity, customer, expansion. Each stage needs exactly one event that moves a record forward, chosen in advance and written somewhere both marketing and sales can see it.

| Stage | The single event that counts | Where teams blur it |
| --- | --- | --- |
| Visitor | A session from a non-bot source | Counting all sessions including brand and support traffic |
| Lead | A form submit with a work email address | Counting chat widget opens and newsletter signups in the same number |
| MQL | Fit score plus one behavioural trigger | Scoring on fit only, so every free email address clears |
| SQL | An account executive accepts within 48 hours | Recording acceptance at the SDR instead of the closer |
| Opportunity | A meeting happened and a next step is diarised | Creating the opportunity when the meeting is booked |
| Customer | Signed contract or first successful charge | Counting verbal commitment at quarter end |
| Expansion | A seat or tier increase on an existing subscription | Mixing expansion into new business ARR |

Two teams can use identical stage names and produce numbers that differ by a factor of five, which is why any funnel review should open by reading the definitions out loud. It takes eight minutes and it regularly ends an argument that has been running for a quarter.

Write the definitions once, keep them in the same document as the numbers, and re-read them at the start of every quarterly review. Definitions drift in whichever direction flatters the person reporting, and a quarterly read-out is the cheapest available control on that.

Notice that half of those blur points inflate the number. That is not an accident. Every stage definition drifts in the direction that makes the reporting team look better, which is precisely why the definition needs a date and an owner, exactly like the other artefacts in the wider [SaaS marketing](/saas-marketing/) operating model.

Creating the opportunity at meeting booked rather than meeting held. Roughly a fifth to a third of booked meetings never happen, so this single choice inflates pipeline, deflates close rate, and makes coverage modelling useless for two quarters until someone finally reconciles the two numbers.

## Why published SaaS conversion rates run from 1.1% to 7.6%

That spread is not a performance gap. It is a counting gap. A site reporting only demo requests against total sessions lands near 1%, and a site counting content downloads, newsletter signups, chat conversations and demo requests against filtered sessions lands near 7%.

Three variables explain nearly all of it. The numerator (which events count as a conversion), the denominator (whether brand, support and existing-customer traffic is stripped out), and the offer (a demo request from a $60K contract value product is a far heavier ask than an email address for a template).

**1.1% to 7.6%** Reported range of B2B SaaS website visitor to lead conversion across published benchmark sets, driven mostly by differing event definitions

Use external benchmarks to check you are not off by an order of magnitude, then stop. Your own trailing six months, measured against a written definition, is the only comparison that will survive scrutiny in a board meeting. The [B2B SaaS marketing funnel guide](/guides/b2b-saas-marketing-funnel/) goes stage by stage on the qualification logic if you need the longer version.

## Stage benchmarks, with the definition attached

These are working ranges for B2B SaaS with contract values roughly between $10K and $60K. Treat them as sanity checks, not targets, and read the definition column before you compare anything.

One number in that table deserves its own paragraph. The no-decision loss is the largest single category of lost deal in most mid market SaaS pipelines, and excluding it from the denominator quietly turns a 19% close rate into a 27% one. If your close rate improved sharply without any change in the sales process, check how closed-lost reasons are being recorded before you celebrate.

## Why the buying group breaks single-lead funnel math

A funnel record is a person. A purchase is a committee. Gartner puts six to ten decision makers in a typical complex B2B software purchase, and in mid market SaaS the working number is usually three to five: the practitioner who will use the tool, a manager who owns the budget line, somebody from security or IT, and occasionally finance.

That mismatch produces three reporting distortions. Lead to opportunity rate looks worse than reality, because five people from one account create five lead records and one opportunity. Attribution looks noisier than it is, because the person who signs is rarely the person who first arrived. And the content plan skews towards the practitioner, since the practitioner is the one who downloads things.

The fix is account level reporting for anything above roughly 15,000 dollars in contract value. Group leads by email domain, report opportunities per engaged account rather than per lead, and track how many distinct people from an account have touched you, because two or more engaged contacts is one of the strongest predictors that a deal will progress. It also changes what you build. A security overview page and a one page business case for a finance reviewer generate almost no leads and quietly unblock deals that would otherwise stall at week six.

## The bowtie: what happens after closed won

The classic funnel stops at the signature, which made sense when software was sold once. In a subscription business, the quarter you book is not the revenue you keep, and a team optimising only the acquisition half can post a strong quarter while net revenue retention slides under 100%.

The [bowtie funnel](/glossary/bowtie-funnel/), from Winning by Design, mirrors the acquisition stages with four post-sale ones: onboarding, adoption, expansion and renewal. Each gets an event, same discipline as before. Onboarding completes at first successful core action. Adoption starts when a second team in the account is active weekly. Expansion is a seat or tier change. Renewal is a signed extension, not a non-cancellation.

At $18K contract value with 110% net revenue retention, a cohort of 50 customers is worth about $900K in year one and roughly $1.09M in year two with zero new acquisition. Push retention to 95% and the same cohort delivers $855K. That $235K swing is larger than most Series A marketing budgets, and content aimed at activation is usually the cheapest lever on it.

Two definitions in that list cause most of the trouble. Onboarding completion usually gets written as an account status rather than a customer action, so it reports 100% while half those customers have never finished the core workflow. Renewal often gets counted as a non-cancellation, which turns an auto-renewing contract nobody has logged into for four months into a success story. Both are the same error as counting a booked meeting: measuring the administrative event instead of the behaviour.

Where the boundary sits between marketing and customer success is a company decision rather than a rule, and it belongs in writing. A workable split gives marketing everything that scales without a human, meaning lifecycle email, in-product education, help centre content and the case studies that arm a champion, while customer success owns anything needing a named person on a call. Arguments about that boundary usually surface the quarter expansion targets appear, and they settle far more easily before then.

Marketing owns more of this half than most teams admit. Onboarding email sequences, in-product education, the help centre articles that rank for feature questions, and the case study that gets a champion promoted all live in the post-sale half and all get built by marketing.

## Product-led funnels, where the trial replaces the MQL

In a self serve motion the MQL and SQL stages usually disappear and are replaced by signup, activation and paid conversion. Activation becomes the stage that predicts revenue, and it is the one most teams fail to define precisely.

The trial to paid number is where benchmark confusion peaks. ChartMogul's analysis across roughly 200 SaaS products found opt-in trials, which require no payment card, converting near 8.9% at the median, while opt-out trials, which capture a card up front, converted near 44%. Both figures are correct and they are not comparable. An opt-out trial has already filtered out everyone unwilling to hand over a card, so it starts with a much smaller and far more committed pool.

| Trial design | Starters per 1,000 visitors | Trial to paid | Paying customers per 1,000 visitors |
| --- | --- | --- | --- |
| Opt-in, no card | 40 | 9% | 3.6 |
| Opt-out, card required | 8 | 44% | 3.5 |
| Reverse trial, full features then downgrade | 45 | 8% | 3.6 |

The customers-per-1,000-visitors column is the one that should drive the decision, and it is remarkable how often three very different trial designs land within a rounding error of each other. Pick on cash flow timing and support load instead: opt-out gets money in sooner and generates more refund requests, opt-in fills the funnel and needs an activation programme behind it.

One further point on trial design rarely makes it into benchmark posts. The conversion rate you publish depends on when you measure it. An opt-out trial measured on day 15, the day after the card is charged, reports a number that falls by roughly a fifth once the refund window closes at day 30. Pick a measurement point, usually 30 days after the first charge, and hold it, because a reporting change that quietly improves a headline metric is exactly the kind of thing that gets discovered during diligence.

Activation is the first action that correlates with month three retention, and you find it by cohorting, not by guessing. For most workflow products it is a second user invited plus one completed core action inside seven days. Write it down, then measure trial to activation and activation to paid separately, because they fail for entirely different reasons.

## Diagnose leaks from the bottom up, always

The instinct when a number misses is to buy traffic. That is the wrong end. Every stage multiplies the waste below it, so a funnel with a 6% close rate that doubles its visitors has simply doubled the cost of the same waste.

**Leak diagnosis order**

There is one honest exception. A brand new product with 400 sessions a month has no statistical basis for any of this, and the right answer there is to go get traffic and stop modelling. Below roughly 2,000 qualified sessions a month, funnel analysis produces noise that looks like insight.

## Funnel math: what the top has to be for ten new customers

Work the chain backwards from the revenue target. This is the calculation that decides whether a plan is ambitious or arithmetically impossible, and it takes about four minutes.

Target: 10 new customers in a quarter at $18K annual contract value, so $180K of new ARR.

| Stage | Conversion assumed | Volume required |
| --- | --- | --- |
| Closed won | 22% of opportunities | 10 customers |
| Opportunities | 50% of SQLs | 45 |
| SQLs | 35% of MQLs | 90 |
| MQLs | 30% of leads | 257 |
| Leads | 2.5% of sessions | 857 |
| Qualified sessions | Top of funnel | 34,280 |

Thirty four thousand qualified sessions a quarter is roughly 11,400 a month, which is a genuinely demanding number for a seed stage company and an unremarkable one at Series B. If that number is out of reach, you have three levers rather than one: raise close rate, raise contract value, or accept fewer customers. Adding traffic is usually the most expensive of the four options and it is almost always the one chosen first.

Two sensitivities are worth running before the plan gets committed. Raising close rate from 22% to 28% cuts required session volume from about 34,300 to roughly 26,900, which is the same as finding 7,400 sessions a quarter for nothing. Raising contract value from 18,000 to 24,000 dollars means seven customers rather than ten for the same revenue, and seven customers needs about 24,000 sessions. Both levers are usually cheaper than buying traffic, and neither one appears anywhere on a funnel diagram.

Plug your own rates into the [SaaS marketing budget calculator](/calculators/marketing-budget/) to see what the same chain costs at your cost per lead, and use the [SaaS marketing budget template](/templates/saas-marketing-budget/) if you need to defend the number line by line. For coverage ratios and the timing question of when pipeline needs to exist relative to the quarter it closes in, [B2B SaaS pipeline math and coverage modelling](/playbooks/b2b-saas-pipeline-coverage/) picks up where this chain stops.

## Cohort by source, or the averages will lie to you

A blended funnel hides the thing you most need to see. A site converting visitors to leads at 2.5% overall is often running 6% on comparison pages and 0.4% on thought leadership, and the average tells you to optimise the site when the real answer is to publish more of one page type.

Cut every stage rate by acquisition source at minimum, and by page type where the data allows. Three cuts pay for themselves immediately. Branded against non-branded search, because branded traffic flatters every rate in the funnel and grows with sales activity rather than marketing effort. Paid against organic, because paid traffic usually converts to lead better and to opportunity worse. First visit against returning, because a large share of B2B conversions happen on a later session, so a first-touch view of the funnel systematically undervalues whatever brought the visitor the first time.

The cost of doing this is real. Segmented reporting needs clean source data, which means UTM discipline nobody enjoys and a monthly audit of unassigned traffic. Budget half a day a month, and accept that 10% to 20% of sessions land in a direct or unassigned bucket no matter how careful you are. That residual is normal, and pretending otherwise produces attribution models that look precise and are not.

## What to report, and what to stop reporting

Report stage-to-stage conversion, volume at each stage, and cost per opportunity by source. That is enough for almost any board conversation about marketing performance.

A harder version of this conversation is now common and worth preparing for. Organic sessions falling while demo requests hold flat reads as a disaster on a session-based funnel report and as a fine quarter on a conversion-based one. Put conversions and opportunities in the headline and sessions in the context column, not the reverse, and the pattern becomes readable instead of alarming.

Stop reporting MQL volume as a headline. It is the easiest number in the company to inflate and the least connected to revenue, and every experienced board member now discounts it on sight. Report accepted opportunities instead, and if sales disputes the source, that dispute is the useful part of the meeting. Attribution arguments get easier once everyone agrees on the events being attributed, which is why [SaaS marketing attribution](/guides/saas-marketing-attribution-models/) is a downstream problem rather than a substitute for stage definitions.

**Publish these definitions next to your funnel numbers**

Two further pieces of context are worth keeping in the same document. Gartner's finding that six to ten decision makers are involved in a typical complex B2B purchase explains why a single lead record rarely represents a single buying decision, and the [ICP template for SaaS](/templates/ideal-customer-profile/) is where the fit half of your qualification logic should come from rather than being invented inside a scoring tool.

## Fix one stage this quarter

Pick the lowest stage that sits outside its band, write the definition next to it, and give one person the number for the next 90 days. One stage, one owner, one definition.

One caution on the ordering. If two stages sit outside their band, fix the lower one first even when the upper one looks worse in percentage terms, because every improvement above a broken stage gets multiplied by the breakage underneath it. A team that lifts visitor to lead from 2% to 3% while close rate sits at 8% has bought itself 50% more wasted sales hours.

Then put the assumed rates into your plan so next quarter's forecast has something to be measured against. The [SaaS marketing plan template](/templates/saas-marketing-plan/) has the stage rows already laid out, and comparing assumption to actual at quarter end is what turns a funnel diagram into a working model of your business.

## Frequently asked questions

### What are the stages of a SaaS marketing funnel?

Visitor, lead, marketing qualified lead, sales qualified lead, opportunity, customer, and expansion. Product-led companies replace the MQL and SQL stages with signup, activation and paid conversion. The stage names matter far less than the event definition attached to each one, because two teams using identical labels routinely count completely different things.

### What is a good conversion rate for a SaaS website?

Published medians for B2B SaaS sites range from about 1.1% to 7.6% of visitors becoming leads, and that spread is a definition problem rather than a performance one. A site counting demo requests only will report near 1%. A site counting newsletter signups, chat opens and content downloads will report near 7%. Compare only against your own trend.

### What is a good trial to paid conversion rate for SaaS?

It depends almost entirely on trial design. ChartMogul's analysis of roughly 200 SaaS products put opt-in trials, where no card is required, near 8.9% median, and opt-out trials, where a card is captured up front, near 44%. Both numbers are accurate. The opt-out trial converts a much smaller and more qualified pool of starters.

### What is the difference between an MQL and an SQL?

An MQL has met a marketing-defined threshold of fit plus behaviour. An SQL is an MQL that sales has accepted and agreed to work. The distinction only has value when both teams agree on the accepting event and the time window, usually acceptance by an account executive within 48 hours rather than an SDR clicking a button.

### What is the bowtie funnel?

The bowtie funnel, from Winning by Design, mirrors the acquisition funnel with a post-sale one covering onboarding, adoption, expansion and renewal. It exists because subscription businesses can post a strong acquisition quarter while net revenue retention falls. If expansion revenue is a material part of your plan, the classic funnel stops measuring too early.

### How do you diagnose a leaky SaaS funnel?

Work from the bottom up. Start at closed won, then opportunity, then SQL, and only fix the top once the stages below are healthy. A funnel with a 6% close rate does not need more traffic, it needs a better qualification bar, and doubling visitors in that state simply doubles the cost of the same waste.

### How many leads do you need for 10 new SaaS customers?

At typical mid market rates, roughly 860 leads. That chain runs 10 customers from 45 opportunities at a 22% close rate, 90 SQLs at 50% opportunity creation, 257 MQLs at 35% acceptance, 857 leads at 30% qualification, and about 34,000 sessions at a 2.5% visitor to lead rate.