How to choose a customer acquisition model

What a customer acquisition model actually is

A customer acquisition model is the path a stranger takes to become a paying customer. The main options are a small set: freemium, free trial, sales-led, and hybrids that combine them. Choosing one is often treated as a matter of preference or fashion — a founder likes the idea of product-led growth (PLG), where the product does the selling, or copies whatever an admired company does. It is not a matter of taste. The right model follows from facts about your product and market, and the wrong one fights those facts every day.

The choice matters because everything downstream inherits from it. Your pricing, your product priorities, whether you hire salespeople, how your marketing works — each is an answer that starts with "given how we acquire customers." Get the model right and the rest of go-to-market has a spine. Get it wrong and you are building a sales team for a product that sells itself, or asking a self-serve product to close six-figure deals with no human in the room.

The three facts the choice turns on

Three facts determine which model fits. Establish them honestly before you compare options, because the models are not really competing on merits — each one requires certain facts to be true, and your job is to find which model your facts allow.

• Time-to-value — how long from first use until a user feels genuine benefit.
• Price and marginal cost — what you charge, and what it costs to serve one more user.
• Buying behavior — how buyers in your market actually buy: alone or by committee, quickly or through procurement.

The first is the most decisive and the most often lied about. Time-to-value decides whether a user will ever reach value on their own. A self-serve motion asks a user to explore alone and feel the payoff before they commit; that only works when the payoff comes fast. If your product takes weeks of configuration to show its worth, no amount of clever onboarding makes a user wait it out alone — they need a human to carry them across the gap.

How the facts map to models

Once you have the three facts, the models sort themselves. Rule out what your product cannot support before you deliberate over what remains.

Model	Requires	Fits when
Freemium	Low marginal cost, short time-to-value, room to upgrade	A free tier can hook users who later need more
Free trial	Value visible within the trial window	The full product proves itself in days
Sales-led	Price high enough to fund a team	Value needs configuration or a buying committee
Hybrid	A proven first motion	Different segments buy in different ways

Freemium gives a limited version free forever and bets that a free tier hooks users who upgrade when they hit its limits. It needs low marginal cost — you are serving free users indefinitely — and a natural reason to upgrade. Free trial gives full access for a fixed window and bets that time pressure forces a decision; it needs value that shows within the window, or the trial ends before the user feels anything. Sales-led puts a human in the motion and needs a price high enough to pay for that human. Hybrid runs more than one of these for different segments.

Price reinforces the mapping. A low price cannot fund a sales team, which pushes you toward self-serve. A high price usually cannot be bought without a human and a committee, which pushes you toward sales-led. When time-to-value and price point at the same model, the choice is easy; when they conflict, the conflict is the real strategic question, and it usually means the pricing or the product needs to change, not that you should force a mismatched motion.

The reverse trial: freemium and free trial, combined

A fourth time-based model sits between freemium and free trial, and it fixes a weakness in each. A reverse trial gives new users the full paid product for a fixed window, then downgrades them to a permanent free tier if they don't convert — instead of locking them out when the clock runs down.

It answers freemium's problem, where a free tier can feel like a scaled-down product that never shows its full value, and the free trial's problem, where the trial ends in a cliff and users who weren't ready simply leave. With a reverse trial the user feels the whole product first, and the fallback keeps them around to convert later.

Notion runs this. New users start on a paid personal tier by default, experience the premium features, and quietly settle onto the free tier if they don't upgrade — no countdown, no cutoff. Linear layers reverse-trial elements on top of a seat-capped free tier. The model needs what freemium needs — low marginal cost to serve the users who never pay — plus a paid experience good enough that losing it is felt.

Motion is a gradient, not a switch

The self-serve-versus-sales-led choice reads like a binary. In practice it is a spectrum, and picking the wrong point on it wastes money either way — a sales team idling on deals a user could close alone, or a self-serve flow abandoning six-figure accounts that needed a human in the room.

Five points span the range:

Motion	Sales involvement	What triggers a human
Pure self-serve	None	Nothing — the user buys alone
Reactive sales	Inbound only	The customer asks to talk
Product-led sales (PLS)	Proactive, on signals	A usage threshold is crossed
Sales-led with self-serve trial	Sales drives; trial is a tool	Sales initiates; the trial is credibility
Pure sales-led	Sales drives everything	Sales initiates; no trial

Product-led sales is the point most modern software lands on: the product acquires and activates users on its own, and a salesperson steps in only when in-product behavior signals a real opportunity — a product-qualified lead (PQL), the behavioral cousin of the older marketing-qualified lead. Datadog and HubSpot run this — self-serve at the low end, a sales team engaging accounts that cross a usage line.

The ends of the spectrum are real models too. Linear sits near reactive sales: mostly self-serve, with a small team answering enterprise inbound rather than chasing it. Snowflake sits at sales-led with a self-serve trial — the trial exists as due-diligence credibility, but the large deals are sold by people from first contact. Where you land on this gradient is a separate decision from the time-axis model, and you make both.

Pricing and acquisition are one decision

Your price is not a separate lever from your acquisition model. The two constrain each other, and the good combinations are specific. A low price cannot fund a sales team, so it points to self-serve; a high price usually needs a human and a committee, so it points to sales-led. That much the mapping above already showed. Pricing structure adds a second constraint on top of price level.

Four structures recur:

Structure	You charge for	Fits
Per-seat	Each user	Products that spread person to person
Per-usage	Consumption — calls, hosts, storage	Products whose value scales with volume
Per-account	The organization, in tiers	Products bought once for a whole team
Per-outcome	A measured result	Products that deliver a discrete unit of work

Match the structure to the motion. Datadog pairs a free trial with per-usage pricing: the trial buys time to wire up data, and once it flows, revenue grows with the customer's infrastructure rather than waiting on an upgrade decision. Notion pairs a reverse trial with per-seat pricing: one person brings it into a team, and seats multiply as adoption spreads. Put per-seat pricing on Datadog and you would be charging engineers for something they don't buy by the seat — the model would fight the product. When the pricing structure and the acquisition model reinforce each other, growth compounds; when they conflict, one of them is wrong.

Why time-to-value is the honesty test

The whole framework rests on an honest read of time-to-value, and this is where teams deceive themselves. Everyone believes their product delivers value quickly. The test is not what you believe; it is what a real user experiences from first login to first genuine payoff, measured or observed, not assumed.

If that read is optimistic, the model built on it fails quietly. A product that actually takes two weeks to show value, launched as a freemium self-serve motion, produces a graveyard of free accounts that signed up, saw nothing, and left. The model was not wrong in the abstract — it was wrong for a time-to-value the team refused to measure. This is why the honest assessment is the human's job and the one part no tool can do for you: the model only works if the read is true.

Start with a wedge, then expand

No product wins a whole market at signup. It wins one narrow use case, then grows from that foothold — the wedge-and-expand pattern. The wedge is the single feature or job your product does so well that a user adopts it without a meeting. Everything else expands from there.

The pattern is visible across successful software:

• Notion entered on notes and personal docs, then expanded to wikis, databases, and project management.
• Figma entered on browser-based design collaboration — the underserved job was making design collaborative, not design itself — then expanded across the organization.
• Datadog entered on infrastructure monitoring, then expanded to logs, APM, security, and more, each product sold to a customer already inside.
• HubSpot made its CRM free as the wedge, then expanded across marketing, sales, and service tiers.

A wedge changes what the acquisition model has to do. It only needs to get one person to value on one use case — a far lower bar than selling the whole platform. That is what makes a self-serve motion viable at all. Once the wedge lands, expansion is a different motion: land-and-expand, where you grow seats, teams, and adjacent products inside an account you already won. Choose the wedge that is easiest to reach value on, not the one with the largest total price tag.

Name the activation event

Signup is not activation. A user who creates an account has done nothing yet; a user who reaches the product's first real payoff has crossed the line that predicts whether they convert. That line is the activation event — a specific behavior, not a page view or a completed tour.

Name it concretely for your product. The clearer it is, the higher your self-serve conversion can climb, because you can design the whole onboarding path to reach one defined moment:

• Datadog — data successfully ingested from a production system.
• Figma — a file created and at least one collaborator invited.
• Snowflake — a real dataset loaded and queried.
• Notion — several linked pages created, the first sign the workspace is being used as one.

Each event is the moment the user feels the value rather than being told about it — the operator's version of the aha moment. A self-serve model lives or dies on how many users reach this event before they drift away, which is why the honest time-to-value read matters so much: time-to-value is just how long it takes a typical user to reach the activation event. Define the event, then shorten the path to it.

Design the anti-persona into the model

Deciding who you will not serve is part of the acquisition model, not a separate exercise. The users a self-serve flow lets in for free cost you something, and the model should quietly turn away the ones who will never convert or never fit.

The sharpest examples build the disqualification into the product itself. Figma makes viewers free and editors paid: anyone can open and comment on a file, but only people doing the paid job — designing — consume paid economics. That single choice does two things at once. It keeps the free-riding persona out of the revenue math, and it turns every shared file into an acquisition channel. Linear does its version through positioning: it is openly built for high-velocity engineering teams, so compliance-heavy organizations that need deep customization and approval gates rule themselves out before they ever start a trial.

Both approaches keep the wrong users from loading up a free tier you pay to run. An acquisition model without an anti-persona invites everyone in and then wonders why the free-to-paid rate is low and the cost to serve is high. Name who the model is not for, and let the design enforce it.

The product becomes the channel

When a product acquires its own users, the product is a distribution channel, not just the thing being sold. Fast-growing software replaced the lead form with a single call to action — "Sign Up Free" — and let people into the product before any conversation. Slack, Asana, and Zoom grew this way, with no lead-generation forms standing between a visitor and the product.

This only works when the product fits the channel. Brian Balfour's product-channel fit names the constraint: a product's growth model has to match the channels that reach its audience, and not every product can grow through every channel. Collaborative, shareable products grow virally, because using them invites others in. Products that solve searchable problems grow through content. Products with high contract value and buying committees grow through sales-assisted trials. A self-serve model bolted onto a product with none of these traits does not become viral by decree — it just quietly fails to spread.

So the acquisition model is also a bet about your channel. Choose a self-serve motion and you are betting the product can carry its own acquisition; choose sales-led and you are staffing the channel with people. Either can be right — but the product has to fit the one you pick.

When to run a hybrid

Hybrids are common and frequently correct. The usual pattern is self-serve at the low end and sales-led for larger accounts — small customers buy themselves while a team pursues the deals worth the cost of a human. This works because different segments genuinely buy in different ways, and forcing them all through one motion serves neither.

The risk is sequencing. Running two motions before you have proven one splits your focus and hides which motion is actually working. Prove the first motion, then add the second when a distinct segment clearly needs it — as a phase-two addition, not a hedge you start with because you could not decide.

Choosing carefully matters because the model is expensive to change once built. Each motion shapes the company around itself: a self-serve model builds a product and marketing team and few salespeople; a sales-led model builds a sales organization and the compensation and process that come with it. Switching motions later means rebuilding those structures, not flipping a setting — which is why the choice deserves the honest analysis up front rather than a guess you intend to correct on the fly. The cost of getting it wrong is not a bad quarter; it is a company organized around the wrong way of selling, discovered after the organization has hardened.

What the recommendation looks like

The deliverable is a freemium/trial/hybrid recommendation: the chosen model, and the three facts that point to it. The rationale is the point. A recommendation that says "freemium, because our time-to-value is under a day, our marginal cost is near zero, and our buyers sign up without a committee" is defensible against its reasoning. One that says "freemium, because it is popular" is a guess wearing a decision's clothes.

Write the reasoning down so the choice can be revisited against evidence rather than relitigated on opinion. If time-to-value later proves longer than you thought, the recommendation tells you exactly which assumption broke — and that is worth more than a confident answer with no visible reasons underneath it.