A SaaS marketing funnel ends in six different places

What a SaaS marketing funnel names

Read three of the published models in a row and something strange happens by the third. Some models describe only the steps before the sale, and others describe only what happens after signup. Both are answering the same query.

The boundary between the marketing funnel and everything after it is rarely drawn at all. One definition draws that boundary explicitly, and it repays reading in full.

The SaaS product funnel is the customer journey that starts after someone signs up for your product, such as during a trial, and continues all the way through conversion, retention, and expansion. Essentially, the product funnel picks up where the traditional sales funnel ends, typically at purchase.

Jinwoo Park, Userflow, The SaaS Product Funnel, 7 May 2025

Once you have that sentence, the field sorts itself. Some publishers cover the span before purchase, some cover the span after signup, and three try to cover both in one list.

The counts differ too, and less interestingly: four, four, four, four, five and six. One model states its position outright and gives the mechanism in the same paragraph.

The six stage version adds activation, retention and expansion after conversion, calling the shape circular because subscription revenue compounds after the sale. That is a claim about payment timing, not about marketing, and it carries the whole argument for running past the purchase.

The stage model itself, the boxes and where they came from, is the same argument as in any other market. That argument is set out in the B2B marketing funnel. What is specific to SaaS is the endpoint.

Why purchase is not the last payment

SaaS pages keep running past the sale for a reason, and the reason is not fashion. Subscription revenue arrives again after the purchase, so a model that stops at the sale stops before the revenue has finished moving. How much it moves is measurable, and it varies by contract size.

SaaS Capital publishes retention medians by average contract value from its 2025 research, on a page dated 18 September 2025. Median net revenue retention runs from 98 percent for companies under $12,000 ACV to 106 percent for those above $250,000. Gross retention runs from 90 to 95 percent across the same six bands.

Retention by contract size, both measures

1. 1Net retention is the blue series. Anything above 100 means existing customers paid more this year than last.
2. 2The rise is not steady. It goes 98, 103, 102, 104, 102, 106, so the relationship holds at the ends and wobbles in between.
3. 3Gross retention, the green series, sits at 90 or 91 percent in every band except the largest. Expansion is doing the work above 100, not lower churn.

The page draws its own conclusion from that chart, that higher net retention correlates with higher ACVs. Read the bars and the correlation holds at the two ends and not in the middle. That is a fair thing to notice about a chart doing otherwise honest work.

What this does not settle is where a funnel should end. A retention number cannot decide the boundary of a model, and reading it that way would be putting more weight on it than it carries. The boundary is a definitional question, and it should be stated explicitly wherever the funnel is drawn.

What the retention figures do settle is smaller and still useful. After the sale, revenue keeps moving, and the direction depends on contract size. Under $12,000 ACV the median company is losing revenue from its existing base. Above $250,000 it is growing without new logos.

Choosing an endpoint you can defend

The choice is not between four stages and six. The choice is whether your model covers one span or two, and who owns the second one. Getting this wrong shows up as a retention target with nobody attached to it.

1. Name the signup event

2. Decide whether marketing owns anything past it

3. Give the second span its own model and its own owner

4. Define activation before you measure it

5. Pick the retention band that matches your ACV

Forrester changes the unit rather than the endpoint. Its B2B Revenue Waterfall asks you to move beyond marketing-qualified leads and to advance opportunities with connected buying groups.

That does not choose your endpoint, and it does narrow the argument. For example, if the thing moving through your funnel is a group, an endpoint defined on one person’s signup is describing a fraction of the decision.

The lead qualification labels sit inside the first span and do not change in SaaS. Where the handoff falls and what each side calls a lead is covered in MQL vs SQL. A product signal can qualify a lead too, and that is a product qualified lead.

What published prices cover

Both spans need software and only one of them has a public price. Marketing automation and CRM publish rate cards. Product analytics and in-app onboarding tools, the things the second span runs on, mostly quote on request.

On HubSpot the entry point is Marketing Hub Starter at $7 a seat a month billed monthly. Professional opens at $800 a month for 2,000 marketing contacts. Enterprise opens at $3,600. Onboarding is a one time $3,000 on Professional and $7,000 on Enterprise. Salesforce prices Sales Cloud per user per month, from $25 on Starter Suite to $550 on Max, with annual billing on the tiers above Starter.

The two vendors' own pricing pages, and the row where no page existsWhat each span costs to enter, where a price is published at all

Every figure comes from the vendor's own pricing page, read 5 September 2026. The last row carries no number because none is published.

Both vendors revise these figures. What they buy is the record of the first span: contacts, deals, stages, reporting.

The second span is the gap, and the last row of that panel is where it shows. Any range quoted here would be invented, so there is none.

One thing worth noticing about the arithmetic: a six stage model and a four stage model cost the same to license. The expense in SaaS is not the diagram. The expense is that the second span needs a team who reads product behaviour, and that is a line most marketing budgets are not already carrying.

Suppose you want the top of your own funnel filled: that is our SaaS SEO service.

Four SaaS marketing funnel errors from a borrowed endpoint

Each of these starts the same way: a stage list copied from a guide whose endpoint does not match the business copying it. The list looks complete, so nobody checks what it covers.

Setting a retention target inside a marketing model. Suppose your model ends at expansion but your marketing team is measured on pipeline. The last two stages then have a target and no owner, so they get reported on every month and worked on never.

Treating the top of the funnel as a queue. The figure that circulates says 95 percent of a target audience is not in the market right now. The underlying work is John Dawes at the Ehrenberg-Bass Institute, and reading it directly beats reading a page that quotes it.

Quoting benchmark bands without a window. Trial to paid at 10 to 25 percent and activation at 25 to 40 percent circulate with no source and no date attached. Bands like these travel fast and mean nothing without a denominator, so they are named here and not used.

Confusing an adoption funnel with a marketing funnel. One published set of four stages describes a single feature: exposed, activated, used and used again. That is a useful model, and if you borrow it for acquisition you will be measuring one feature and calling it your funnel.

Which window retention arithmetic needs

Two numbers matter in your second span and both are easy to compute wrongly. Net revenue retention compares a cohort against itself over time, so fix your cohort and your period before you read anything into it.

Net revenue retention compares the recurring revenue from one set of customers at the start of a period against the revenue from those same customers at the end. Upgrades, downgrades and churn count; new customers do not. SaaS Capital reports medians from 98 to 106 percent depending on contract value.

The loss underneath retention has its own vocabulary, and a SaaS churn rate means one thing when you count customers and another when you count revenue.

Gross retention leaves out the upgrades. Across the same six bands it sits between 90 and 95 percent, so it shows how much of the base holds without expansion covering for it. Every band except the largest lands at 90 or 91. Most of the spread between a 98 and a 106 comes from the base holding: gross retention moves five points across the bands, and expansion accounts for the other three.

The first span keeps its own number, and it is small. First Page Sage puts visitor to conversion at 1.1 percent for B2B SaaS, from client data gathered between January 2022 and August 2025. At that rate a hundred conversions takes roughly nine thousand visitors.

A vendor funnel guide quotes the same figure, and the citation holds up when you follow it back to First Page Sage.

Trial and freemium are not the same entry either, and the distinction usually gets blurred. A trial has an end date, so the decision arrives on a schedule you set. A freemium plan has no end date, so the decision arrives when the customer hits a limit, or never. The same activation rate means different things under each.

That difference lands directly on the first span. A trial gives marketing a deadline to work against. A freemium plan does not, and the work moves inside the product where marketing usually has no surface. Choosing between them is a funnel decision dressed up as a pricing decision.

Activation is the stage that never gets defined. Wherever it is named as a stage, the denominator and the window are left open. Until you write yours down, an activation rate is a number that moves when the definition moves, and nobody will be able to tell the two apart.

The same open definition problem sits in the layer above. For example, whether you can compare your retention number against anyone else's depends on which SaaS metrics feed which.