Customer Centric Solutions LLC

Demand, diagnosed

The Funnel Is a Model of One Purchase

A funnel can describe how one customer arrived. It has no word for the thing that made the next one arrive.

John Gusiff · 23 August 2026 · 6 min read

You probably met Notion through a page somebody else made. A team wiki a colleague shared. A content calendar template on YouTube. A public roadmap from a company you were researching. The first thing you saw was not Notion. It was another person's work, sitting inside Notion.

Ask a growth team to write that down and it lands in a row labeled referral, or organic, or community. All three are channel names. A channel is a place customers arrive from. None of those rows records the fact that actually explains the arrival: the artifact that brought you in was manufactured by the previous customer, out of the product itself, as a byproduct of using it.

That fact has nowhere to live in a funnel. Not because the funnel is wrong. Because it is a model of something else.

What a funnel is a model of

A funnel models one acquisition event. Awareness, consideration, conversion, in whatever local vocabulary a company prefers. Its unit of analysis is a single buyer moving through a single sequence, once. Within that scope it is a good model, and an unusually honest one. It produces rates, leaks, and a cost per stage, all of which are checkable against reality. Most strategy artifacts cannot claim as much.

Its limit is structural rather than moral. Every cohort enters at zero. Nothing that happened to the last customer changes what the next one costs. Run a funnel well for three years and the top of it costs roughly what it cost on day one, adjusted upward for auction inflation. There is no cell in the model where accumulation could be recorded, because accumulation is not the thing the model was built to describe.

The old model: a funnel A funnel narrowing left to right along a horizontal flow, from a trigger through four stages to conversion, losing volume at every stage. A dotted return path shows the next cohort starting over. The two panels below, compounding asset and retained value, are empty: the model has no cell for either. THE OLD MODEL One acquisition event, modeled end to end. Nothing carries from this customer to the next. ALL DEMAND ONE PURCHASE TRIGGER Demand shows up Awareness Consideration Evaluation Decision Conversion VOLUME LEAVES AT EVERY STAGE NEXT COHORT ENTERS AT ZERO WHAT THIS MODEL HAS NO CELL FOR COMPOUNDING ASSET ( empty ) RETAINED VALUE ( empty )
Figure 1 · The funnel narrows along a single horizontal pass, from all demand to one purchase, shedding volume at every stage. The two slots underneath stay empty by construction.

Two stocks

A loop is not a funnel bent into a circle. The test is whether two things happen at once.

Compounding Asset

Something accumulates on the business side.

The thing the company keeps, which grows with use and makes the next cycle cheaper, faster, or better than the last one.

Retained Value

Something stays with the customer.

What the customer still has after the session ends, stated in the customer's own words. If it reads like a tagline, it is copy, not retained value.

Both cells have to fill. Business-side compounding with nothing retained on the customer side is extraction, and it churns. Customer-side retention with nothing accumulating on the business side is a good product with a flat cost curve, which is a fine thing to own and not a loop.

Run it, do not label it

Notion appears on every product-led growth list ever published. Product-led implies a User loop: the product spreads itself, users pull in users, invitations do the work.

Diagnostic step · The Fuel Test

Trace the fuel instead of reading the label.

Do not ask what motion the company is filed under. Ask what physically carries the product into the next context, and whether that thing was made by the last customer. The label is a claim. The fuel is evidence.

Run the Fuel Test on your own model

What spreads Notion is an accumulating library of user-made templates, workspaces, and public pages, found through search, YouTube, Reddit, and TikTok. Notion let creators sell those templates and took no cut, which converted a slice of its user base into a distribution tier it does not pay for and does not manage. Product invitations exist, but they ride underneath.

The tell: each one converts off a template, having met no one. The template does the acquiring.

Product manager

Finds a product-roadmap template through search. Adopts Notion.

Startup team

Discovers a startup operating-system template through a creator. Adopts Notion.

Designer

Finds a portfolio template on a creator's site. Adopts Notion.

The label said User. The fuel says Content.

This is the discipline the funnel era never required, because in a funnel the label and the mechanism are the same thing. In a loop they come apart, routinely, and the label is usually the more flattering of the two.

Lightning lesson

This piece stops at funnel versus loop. Which loops you should expect to find, given your primary GTM motion, is the subject of the lightning lesson Loop or Funnel? Diagnosing Your GTM Motion, run with Krzysztof Czubak. It walks seven motions and the loop stack each one tends to sit on, then runs the fuel test live on this Notion case.

Watch the lesson

Run the two stocks on Notion properly and three loops come back, coupled.

The new model: three coupled loops Notion runs a dominant Content loop for acquisition, a supporting Skill loop for retention, and a subordinate User loop for expansion. Each loop names a trigger, four user actions, a compounding asset, and a retained value. Arrows between the loops show each loop feeding the others. THE NEW MODEL Three closed loops, each feeding the others. The shape Notion uses to compound its own growth. TRIGGER Need to structure or improve recurring work SKILL LOOP RETENTION · SUPPORTING Build workflow Use workflow Learn what works Refine and reuse COMPOUNDING ASSET Team proficiency + reusable working patterns RETAINED VALUE Faster, easier, more consistent work TRIGGER Need to capture or organize work CONTENT LOOP ACQUISITION · DOMINANT Create page Organize content Link knowledge Reuse or share COMPOUNDING ASSET Connected workspace content RETAINED VALUE Reusable organizational knowledge TRIGGER Need another person to participate USER LOOP EXPANSION · SUBORDINATE Share workspace Invite collaborator Collaborate Create and share COMPOUNDING ASSET Active collaboration network RETAINED VALUE Connected people, context, and work Reusable patterns improve content Connected knowledge improves workflows Collaborators create more content Shared work creates reasons to invite
Figure 2 · Content is dominant and does acquisition. Skill supports and does retention. User is subordinate and does expansion. Each loop's output arrives as another loop's trigger.

The Content loop is dominant and does acquisition. Trigger: the need to capture or organize work. The user creates a page, organizes content, links knowledge, reuses or shares it. Compounding asset: connected workspace content. Retained value: reusable organizational knowledge.

The Skill loop supports and does retention. Trigger: the need to structure recurring work. The user builds a workflow, uses it, learns what works, refines and reuses. Compounding asset: team proficiency and reusable working patterns. Retained value: faster, easier, more consistent work.

The User loop is subordinate and does expansion. Trigger: the need for another person to participate. The user shares a workspace, invites a collaborator, they work together and create more shared material. Compounding asset: an active collaboration network. Retained value: connected people, context, and work.

The seams matter as much as the loops. Reusable patterns improve the content. Connected knowledge improves the workflows. Shared work creates reasons to invite. Collaborators create more content. Each loop's output arrives as another loop's trigger, which is why "channel mix" is the wrong description of what is happening here. A mix is a set of independent inputs. This is one system with three entry points.

Channel or asset

A funnel would file the template gallery under channels, and the filing would feel correct, because traffic arrives through it. The distinction that matters is ownership and slope.

Channel

Rented and flat.

Stop paying and it stops producing. It costs the same next quarter as it did this one.

Asset

Owned and sloped.

It was built once, by someone else, and it is still working.

The test takes one question. Remove the artifact and does acquisition stop? Notion without user-created pages is a blank editor with an advertising budget.

Where three loops still leave you exposed

Three coupled loops is more legs than most companies have. It is not immunity, and the pressure points are visible in the diagram itself.

The first sits at the Content loop's trigger. Its entry condition is exposure to somebody else's artifact worth copying. Copying is a workaround for the cost of construction. When a model will construct the structure on request, that cost falls, and the trigger weakens with it. The gallery does not disappear. It stops being the front door, which is a different and more serious problem, because the front door was the dominant loop.

The second sits at the Skill loop's retained value: faster, easier, more consistent work. Read it as a question rather than a claim. Do users keep genuine competence, or sunk-cost inertia dressed as loyalty? That question was always the wall on this loop. Under AI it gets sharper, because if an agent builds and maintains the structure, proficiency accumulates in the agent rather than in the team. Retained value an agent can absorb was a rental all along.

Notice which loop is least exposed. It is the User loop, currently the subordinate one. Coordination among people is not a generation task, and a network of colleagues who work in the same place is not something a model produces on request. Whether the subordinate loop stays subordinate is the live strategic question, and it is a question you can only ask once the three loops are on the page.

The question to take to Monday

Name one thing a customer did last month that is still acquiring customers this month. Not a campaign. An artifact, made by a customer, out of your product, still working without supervision.

If nothing qualifies, the growth is a funnel, whatever the deck says. That is not a verdict against the business. Funnels convert demand, and converting demand is most of what most companies do. It is a verdict against the vocabulary. Ask a model of one purchase to explain a system, and it will answer in channels, because channels are the only words it has.