Revenue is a system, not a funnel

Do not index

The Predictive Path

Course 2: Revenue as a system

Lesson 1: Revenue is a system, not a funnel

Revenue is a system, not a funnel

Why the funnel became every company's model of revenue, where it quietly stops working, and what the customer's full life shows instead.

The funnel is the most comfortable picture in revenue — and a genuinely useful one, for a step. Wherever prospects or customers drop off — a trial, a sales stage, a renewal — a funnel models that stage well. The mistake is scaling it up to stand for the whole: revenue isn't one long funnel from lead to renewal, it's a system with funnels inside it, and the outcomes that decide the year — expansion, cost, margin, churn — live in the parts a funnel can't see. This lesson is about what you stop being able to see the moment you treat the whole of revenue as a funnel instead of a system.

Why the funnel became the model

Almost every revenue organization pictures itself as a funnel. Prospects enter at the top, move through stages, a fraction converts, revenue comes out the bottom. It is the first diagram drawn in most planning meetings and the shape most leaders reason in without noticing they have chosen a shape at all.

It earned that position. The funnel fit the world it was built for. When growth was dominated by new customers, when sales motions were relatively uniform, when markets moved slowly enough that this year resembled last year, the funnel was an accurate enough model of how revenue actually formed. It gave teams a shared language. It gave executives a way to see where deals stalled and a number to manage toward. For a business whose growth really did come mostly from new logos moving through a defined sales process, the funnel was not a simplification — it was a reasonable map of the territory.

The funnel did not become universal by accident or by laziness. It became universal because, for a long time and for many companies, it worked. That is worth saying plainly, because the rest of this lesson argues it has stopped being sufficient — and that argument only matters if the funnel was genuinely good to begin with. It was. The question is whether the world it fit is the world the company is now in.

Where the funnel quietly stops working

The funnel has one defining property, and every limitation follows from it: the funnel ends at the close. It is a model of throughput toward a signed deal. Once a deal reaches the bottom, the model is finished with that customer — attention returns to the top.

In a recurring-revenue business, that property is the problem. Consider what actually decides whether a customer was worth acquiring: whether it renews, whether it expands or quietly contracts, what it costs to serve and whether that cost rises, when the contracted revenue becomes collected cash, whether the margin survives the second and third year. None of those is a throughput question, and not one is decided before the close. Every one is decided after it — on the part of the customer's life the funnel has already stopped watching.

This is not a flaw more stages would fix. Adding post-sale stages produces a longer funnel, and the thing that was missing was never length. What is missing is that the funnel has no way to represent a consequence that forms after the close from a decision made before it. A model whose shape says "the work ends at signature" cannot describe a business whose economics are mostly determined after signature. The funnel ends precisely where recurring revenue begins, treating the close as the finish line when in this business the close is much nearer the start.

Course 1 ended on exactly this gap, seen from the customer's side: a single customer that was four facts — demand, deal, usage, margin — each settled at a different point in its life, read by a different function, never assembled. The funnel is why that customer looked fine at every point anyone looked. It was watching the one stretch where it was fine and is silent on the stretches where it was not, because those stretches are not in the model.

What the customer's full life shows instead

Follow one customer along its whole life rather than only through the deal. It is acquired from a segment at a cost, signs a contract with a structure, adopts the product or does not, expands or contracts, costs something to serve that moves over the relationship, and renews or churns — and what it turned out to be is information about what the next acquisition should look like, if anything carries that back. Call that full path the lifecycle. The funnel is a short early section of it.

Along that path, the decisions interact, and the interactions decide the outcome. Three, kept short:

The segment marketing acquired from sets a cost-to-serve customer success absorbs a year later — invisible in the funnel, which saw only an efficient acquisition.

The contract terms sales gave to close move the cash timing and second-year margin finance reports well afterward.

The expansion customer success drove lands in accounts that were already expensive to serve, changing a margin no one forecast.

None of those effects lives in the function that caused it. Each is a relationship between a decision at one point in the customer's life and a consequence at another, owned by someone else — and a funnel, being stages in a line with no connections between them, has no representation for any of it.

Revenue enters from more than one place

There is a second thing the funnel flattens, and in a modern SaaS business it is the larger one.

The funnel implies revenue has one source: new prospects entering the top. That was roughly true when growth was new-logo growth. It is not true now. Revenue arrives from several distinct streams — new-logo acquisition, expansion within existing accounts, upsell into higher tiers, usage-based growth that accrues without a deal, partner-sourced revenue, and revenue recovered by reducing churn. A healthy company runs most of these at once, and in many the new-logo stream is no longer the largest.

Each of those streams enters the same lifecycle structure — acquisition, contract, adoption, cost-to-serve, expansion or contraction, renewal — but travels it differently. A usage-based customer has almost no acquisition cost and a long, quiet adoption curve that determines everything. An expansion has no acquisition cost at all and a cost-to-serve already in motion. A partner-sourced customer carries a margin shaped by the partner economics before it ever signs. A new logo carries the full acquisition cost the funnel was built to track. Same lifecycle shape; different cost to enter it, different onboarding path, different expansion behaviour, different cost to serve, different point at which it becomes profitable.

So the real picture is not one funnel and not even one lifecycle. It is many lifecycle lines running at once, fed from several different streams, each stream behaving differently along the same structure, with the company's result being whatever that whole population nets out to. The funnel models the entry of one stream and is silent on the existence of the others.

So revenue is a system

Put the history and the comparison together and the conclusion is forced rather than asserted.

The funnel was a good model of a business whose growth came through one early stretch, fed by one stream. The business it is used for now has its outcomes decided across the whole customer life — renewal, expansion, cost-to-serve, cash timing, margin — fed by several streams that each travel that life differently and interact as they go. A model that ends at the close and assumes one source of revenue cannot hold any of that, and no effort inside the model recovers what the model has no representation for.

What the business actually is, then, is a population of customer lives running at once, entered from several different streams, interacting along their length, with the company's results being what that population produces in aggregate. That is what the word system means here, used plainly: a set of parts whose behaviour is determined by how they connect over time and how they are fed, not by what any one of them does alone. The funnel is one accurate instrument for the entry of one stream into one early stretch. It was never the system. It was always a short section of it.

This lesson changes the object in view and stops there, deliberately. It does not yet say how the parts should be connected, what should be carried back from the end of a customer's life to the start of the next, or how a system like this is steered rather than watched. Those are the rest of this course. Reaching for them now would repeat the mistake Course 1 spent five lessons diagnosing — taking the fix before the object is fully in view. The funnel keeps its place: still the right model for the deal stretch of the new-logo stream, no longer mistaken for the whole.

Next up

If revenue is a system, improving one part of it is not automatically good news.

→ Continue to Local optimization breaks the system

———

This article is part of The Predictive Path

By Niko Laine, SaaS CFO

→ LinkedIn

You can read each lesson on its own or explore the full curriculum.

→ Explore courses

→ About