How to Price

This is a narrative theme with a typological spine where sources diverge, synthesising what Patrick Campbell, Madhavan Ramanujam, Naomi Ionita, Eoghan McCabe, Bret Taylor, Albert Cheng, and Julia Schottenstein say about pricing as a discipline — drawing also on Monetizing Innovation.

The core claim

Pricing is the most under-worked lever in most companies. Naomi Ionita’s research finding — a 1% improvement in monetisation has 4× the bottom-line impact of a 1% improvement in acquisition — follows from simple arithmetic: monetisation acts on the full existing user base; acquisition acts only on the marginal new user. Yet most product and growth teams over-invest in acquisition because it is measurable and culturally exciting, and treat pricing as a one-time decision made at launch, rarely revisited. Patrick Campbell found that fewer than 10% of SaaS companies revisit pricing quarterly.

The recurring discipline across sources is to charge on a value metric — the unit that scales with the value the customer receives, not with the vendor’s costs. Get the value metric right and the rest of pricing can be mediocre without fatal harm; get it wrong and pricing fights the business permanently. Madhavan Ramanujam’s complementary prior: have the willingness-to-pay conversation before building, not after. Most innovation failures (he cites 72%) are pricing failures disguised as product failures.

Before the product: willingness to pay as design input

Madhavan Ramanujam and Georg Tacke’s Monetizing Innovation established the central methodological point: price is a measure of value the way a litre is a measure of volume. Testing it at the idea stage is simply test-and-learn applied to monetisation. Companies that skip this step do not discover they have a pricing problem at launch — they discover they have a product problem, because the product was built around features rather than around what customers will pay for.

Julia Schottenstein applies the same logic operationally at dbt Labs: ‘you don’t get to decide if you’ll have a willingness-to-pay conversation — only what.’ dbt Labs’ early practice was to charge a small fraction of delivered value (value creation over value capture), then run a first-ever price increase at low stakes, purely to learn elasticity before it mattered.

Ramanujam’s methods for surfacing willingness to pay — relative anchoring, acceptable/expensive/prohibitively-expensive questions, trade-off exercises — sit upstream of any packaging or price-point decision. They answer ‘for whom, and for what’ before ‘how much.‘

The value metric: the dimension of charge

The value metric is what you charge on: per seat, per thousand API calls, per invoice sent, per video, per resolved ticket. Patrick Campbell ranks it the single highest-leverage pricing decision a company makes, ahead of headline price, packaging, and discounting.

A well-chosen value metric works on all three growth levers simultaneously.

Acquisition. Price scales with customer value automatically. A large enterprise pays enterprise prices and a two-person startup pays startup prices, even when their raw unit consumption is similar, because the metric tracks value derived rather than units used.
Retention. Churn runs 20–25% lower. Customers facing a budget contraction downgrade along the metric rather than cancelling, avoiding the ‘I’m paying for far more than I use’ resentment that drives outright churn.
Expansion. Expansion revenue roughly doubles. Instead of pitching a higher tier’s features, the business observes usage and moves customers up automatically.

Naomi Ionita reaches the same conclusion from the growth angle: usage-based pricing (API calls, documents created, invoices sent) creates a natural escalator that aligns the vendor’s incentive with the customer’s growth. Seat-based pricing, the SaaS legacy default, misaligns them.

Madhavan Ramanujam distinguishes the metric from the headline price with the Michelin tyre example: charging per mile rather than per tyre changed customer behaviour and perceived value far more than any price-point adjustment could have.

How you charge vs how much you charge

The sources broadly agree that the structure of pricing matters more than the level.

Ramanujam’s behavioural pricing toolkit — decoy pricing, the compromise effect (good/better/best packaging), pennies-a-day framing, the Panini effect (puzzle-completion compulsion) — can shift revenue mix and ARPU substantially with zero product changes. One SaaS company he cites achieved 30%+ MRR growth purely by reframing three-tier packaging.

Naomi Ionita’s diagnostic process reflects this priority. Her recommended pricing committee — cross-functional, including product, growth, sales, and finance — works through the Van Westendorp method (four survey questions identifying the price-too-cheap, good-value, expensive-but-acceptable, and too-expensive thresholds) before touching headline numbers. The Envoy story she recounts makes the upside vivid: Larry Gadea charged ten times his normal price in an enterprise call, on instinct, and the prospect accepted without hesitation. Most companies are underpriced, and enterprise sales conversations are the fastest way to find the ceiling.

Patrick Campbell’s tactical advice when internal politics make the value-metric conversation impossible is to start elsewhere — with a straightforward annual price increase — to build the data and enablement muscles, then return. Either way, the discipline is to track one KPI (revenue per customer) and push it up every quarter through a small pricing committee on a recurring cadence.

Freemium: acquisition channel, not a default

The sources converge on a sharp position: freemium is a deliberate acquisition strategy, not what you do when you have not thought about monetisation.

Naomi Ionita’s freemium calibration principle: the free tier should demonstrate value clearly enough to create desire, and create a genuine capability limit clearly enough to motivate conversion. Both conditions are required; either alone is insufficient. Evernote’s failure is her canonical cautionary tale. The free tier was capable enough that most users had no material reason to upgrade. One survey finding crystallises the problem: ‘I feel guilty’ was a leading purchase driver. If guilt motivates conversion, the free version is too good and the paid version is not differentiated enough.

Albert Cheng’s Grammarly experiment runs in the same direction from the other side. Free users only ever saw spelling and grammar fixes — the floor of the product — and so perceived Grammarly as narrower than it was. Intermingling a limited number of premium suggestions into the free user experience nearly doubled upgrade rates. His principle: the free product should reflect the full value of the offering, acting as a reverse free trial woven into real-time usage.

The structural distinction Naomi Ionita draws: freemium that works builds a conversion funnel; freemium that fails builds a substitute. Patrick Campbell routes free users into a ‘pool’ via inbound media, not as a permanent state but as a middle-of-funnel step toward a buying conversation.

Where sources diverge: seat/usage pricing vs outcome-based pricing

The largest fault line in the cluster is between value-metric pricing (charging on a usage or seat dimension that correlates with value) and outcome-based pricing (charging per achieved result, bypassing the consumption proxy entirely).

Value-metric pricing assumes consumption is a reliable proxy for value — more API calls, more invoices, more seats means more value. This holds for most SaaS. But it breaks for AI agents, where the agent’s value is in what it accomplishes, not in the compute it consumes.

Eoghan McCabe built Intercom’s Fin pricing on this distinction. Old pricing was ‘complex, multi-metric, widely resented.’ New model: $0.99 per resolved customer service ticket, against the $20–30 per ticket that B2B companies were spending on human-staffed customer experience. Early unit economics were inverted — it cost $1.20 to deliver each $0.99 resolution — a deliberate bet on LLM cost curves improving. The underlying principle: price from value, not cost. ‘The cost is our problem.’

Bret Taylor at Sierra makes the structural argument explicit. Outcomes-Based Pricing is only viable when outcomes are measurable and containment rates are high enough — below roughly 50% agent resolution rate, the unit economics collapse; above 90%, customers may question whether the price has become too expensive relative to human labour. But when those conditions hold, outcome-based pricing captures 25–50% of value created, versus the 10–20% typical of SaaS.

Madhavan Ramanujam’s attribution–autonomy two-by-two maps where each model fits. High autonomy and high attribution (fully agentic AI with measurable outputs) is the outcome-based quadrant. Low autonomy and low attribution (AI as copilot, human decision-making retained) is still a usage or subscription quadrant. The error is applying the outcome model before the attribution condition is met — which is why, he notes, only around 5% of AI companies have reached it today.

The retention side of the pricing decision

Pricing decisions feed directly into retention, and the sources treat this link as non-negotiable.

Patrick Campbell’s Tactical vs Strategic Retention distinction is the operational entry point. Strategic retention is product work — ICPs, time-to-value, features. Tactical retention is the mechanics: payment-failure recovery, cancellation flows, offboarding, pause plans. Past product-market fit, tactical retention accounts for 25–40% of churn, but product teams systematically neglect it. A well-chosen value metric reduces the churn problem at the strategic level before tactical retention is needed — customers downgrade rather than cancel, removing the all-or-nothing binary that generates churn.

The Modern Growth Stack situates monetisation within the full growth hierarchy. Acquisition is the most resourced layer in most companies; monetisation is the most neglected. The 4× impact ratio is the arithmetic case for rebalancing. See also The Anatomy of Growth for how monetisation fits as one lever among the full growth system — this theme covers it as the whole subject.