True Value

Features get copied in a weekend. Compute gets cheaper by the month. The only thing you still own is what your customer believes you're worth.

Drive up Route 1 North through Wiscasset, the self-appointed prettiest village in Maine, and you pass a family hardware store called Ames True Value. This is Vacationland, where the plates promise “the way life should be.” Ask the locals why they love it and few mention the name over the door. They mention the staff who remember them and the honest prices, the wood preserver they found online for forty dollars and bought at Ames for thirty-two. The value in True Value has little to do with the two words on the sign. It lives in what the customer experiences and believes. [10]

Hold that thought. In 1776 the same idea drove Adam Smith slightly mad.

The paradox that will not die

Smith noticed something that annoyed economists for a century. Water keeps you alive and costs almost nothing. Diamonds are useless, yet they cost a fortune. Nothing about the thing itself explains the price. He called it the paradox of value, then, being Smith, shrugged and moved on. [1]

It took until the 1870s for Carl Menger and the marginalists to crack it open, and their answer is still uncomfortable. Value is not a property of the object; it is a judgement the buyer makes, at the margin, in a context. A glass of water is worth nothing next to a tap and everything in a desert. The diamond has not changed; the perceiver has. Oscar Wilde put it more rudely: a cynic knows the price of everything and the value of nothing. [2] Price is public; value is private and perceived. They are not the same number, and the gap between them is where your commercial future lives.

Which brings us to your (AI) product. If you own a number, a pricing page, a packaging call or a renewal, closing that gap is your job.

The intrinsic-value fallacy

Every technical founder carries the same bias, and Martin Casado of a16z named it: you build something clever and assume it is intrinsically worth a lot of money because it is intrinsically clever. It is not. Ben Horowitz warned Casado that no single decision will impact your company's valuation more than pricing, and the reason is precisely Menger's point. [3] In an early or disrupted market, the customer has no idea what your thing is worth; they value it by how you present and price it. Enter cheap because “we will monetize later,” and you have taught the market that your water is, in fact, water.

This has never mattered more than right now, in the middle of the so-called Saaspocalypse.

Why the old moat leaks

For twenty years, SaaS sold access. Pay per seat, log in, book the ARR, enjoy gross margins near ninety percent. That model is quietly dying, and three things are draining it at once. First, AI carries real variable cost, so seat pricing that ignores usage bleeds margin as adoption grows. Second, features are no longer defensible: what once took a moat to build now takes a weekend to wrap around someone else's model. Third, credit and token costs keep falling, so any value priced on top of raw compute erodes underneath you.

The industry's answer is to march rightward along the pricing spectrum, seat to usage to work to outcome. Useful, but notice the trap. Gross margins fall as you go right, from around ninety percent on subscription toward thirty percent on pure work-based models, because you are now reselling variable cost. [7] Usage-based pricing on its own just relocates the margin problem; it does not solve it.

So what is the moat when features are copyable and compute is a commodity? The one thing a competitor cannot clone in a weekend: the value your customer perceives, believes, and can point to on their own P&L. Perceived value is the last defensible layer; everything below it, the model, the interface, the credits, is rented. Here is how to raise it, in three levers you can start moving this quarter.

Lever one: charge for the work, not the login

Menger again: people value the outcome, not the mechanism. Nobody at a hardware store wants a drill; they want a hole. So charge for what your AI does, tickets resolved, workflows finished, leads qualified, not for the right to open the app.

The operator move: pick one usage-linked value metric per product, one that climbs when the customer succeeds, not when they add logins. Model it against three or four live accounts before you publish. Expansion then stops being a fresh sale and becomes a meter reading: as usage climbs, revenue climbs, and NRR does work your reps used to chase.

The best AI companies never show the customer their rate card. Credits, tokens and API calls are your cost language, not the customer's. In the QBR deck, lead with “3,000 tickets resolved” and “40 hours saved,” and keep credits and tokens off the slide. Kyle Poyar of Growth Unhinged calls this the abstraction layer. [4] It protects margin when vendor prices shift and keeps the conversation on value, not your bill of materials. The moment a customer sees your ingredient costs, they price your ingredients. Do not let them.

The customer sees outcomes; credits and tokens stay off the invoice. Source: Kyle Poyar, Growth Unhinged [4].

Lever two: package so the value is legible

Good packaging is not decoration; it is how you make value visible before it is billable. Gibson Biddle's DHM test is the right filter for every feature and tier: does it Delight the customer, in a Hard-to-copy, Margin-enhancing way? [5] If it delights but is trivially copied, it is table stakes, not a tier. If it enhances margin but nobody notices, that is a communication problem, not a product.

Score each capability against the three levers. Three ticks means build with conviction; a cross is the trade-off to weigh. Source: The Principles of Pricing, Notion Capital [8]

The workhorse structure is still Good, Better, Best, because a customer can rank it in five seconds. A business-class seat and an economy seat fly to the same city, yet everyone understands why one costs more. [8] But Good, Better, Best is only one shape; pick the structure that matches how your customers actually buy.

Six common ways to bundle and sell features. Good, Better, Best is the default, not the only option. Source: The Principles of Pricing, Notion Capital [8].

Here is the operator move. List every feature, then score each on two axes, expected adoption and willingness to pay, and let the grid sort them. High adoption and low willingness to pay goes in the base plan, so it feels generous. High adoption and high willingness to pay anchors your middle and top tiers. Low adoption and high willingness to pay becomes an add-on. The rest gets cut.

High-value features monetize as add-ons or premium; high-adoption low-value features become standard; the rest get cut. Source: The Principles of Pricing, Notion Capital [8].

Run it as a ninety-minute workshop with product, sales and CS in the same room, because each sees a different slice of the truth. Add-ons double as a land-and-expand lever: ship the new capability as a paid add-on first, watch the attach rate, then fold the winners into the tiers later.

Lever three: prove it, or it is not real

Now let's be blunt. If you cannot measure the value, you cannot price it. And if you cannot show the customer the measurement, they will not believe it. “We just know our top customers love the ROI” is not evidence; it is a feeling. If it is not substantiated with data, it is not true. It is a hope with a slide deck.

The market gap proves the point. Roughly ninety-two percent of companies now use AI in operations, only about five percent extensively, and nearly half report no measurable impact. [7] Half the market is getting value they cannot see, so they cannot justify the spend and churn. That is not a product failure; it is a perceived-value failure, and the biggest opportunity on the table. Build the measurement into the product and onboarding, and you win the renewal before it is contested.

Concretely, three artifacts earn their keep. First, a value calculator on the pricing page where a prospect plugs in their own numbers and sees the result; self-quantified value converts. Second, the same calculator in your reps' hands during discovery, so “why now” carries a euro figure. Third, a realized-value review ninety days after close that turns promised ROI into a measured case study, then feeds the next deal. [3] Before you set the number, run a Van Westendorp survey per segment to bracket your floor, anchor and ceiling. [6] Then instrument one metric per account and surface it in every renewal.

Let the prospect quantify the value themselves; self-served numbers convert. Source: Commercieel Verbeteren, AI Pricing Playbook [9].

The moat, restated

Put the three levers together and the moat reappears somewhere new. A competitor can copy your feature. What they cannot copy is a customer who has watched forty hours a month vanish from their support queue, has your outcome baked into their P&L, and climbed your tiers until leaving got expensive. Features are rented. Compute is rented. Perceived, proven, embedded value is owned, and it compounds every quarter you demonstrate it.

That is the whole game in a Saaspocalypse. The question is not “how much is my technology intrinsically worth,” since Menger settled that in 1871: nothing, on its own. It is “how high can I raise what my customer perceives, and how durably can I prove it.” Price is what you put on the page; value is what they believe. True Value is the second number, and the only one you can control.

Which is why a little hardware store in Wiscasset outlasts the big boxes. It sells the same nails as everyone else at a fair price, but the value people walk out with is trust, competence and the sense that someone remembered them. That is a moat you cannot buy in bulk. It might even be the way life should be. [10].

So what should you do on Monday?

Price the outcome, hide the ingredients. Put a work-based line item on the order form and keep credits and tokens off the customer's slide. [4]
Package for legibility. Score every feature on adoption and willingness to pay, then build Good, Better, Best and add-ons from the grid. If value is not visible, it is not valued. [8]
Prove it or drop it. Ship a value calculator, run Van Westendorp per segment, and review realized value ninety days after every close. A feeling is not a number. [6]
Climb the maturity ladder and bring the customer's perception with you. The higher you go, the more you can charge, but only if they can see it. [9]
Price early, price often, and never confuse the price of your product with its value. [3]

Nobody cares about your features. Stop shipping, marketing and selling features. Start proving value. That's the only moat left

Sources and credits
1. Adam Smith, the paradox of value (The Wealth of Nations, 1776), and Carl Menger, the subjective / marginal theory of value (1871). Classical economics, public domain.

2. Oscar Wilde, “a man who knows the price of everything and the value of nothing” (Lady Windermere’s Fan, 1892). Public domain.

3. Martin Casado, Ben Horowitz, Mark Cranney and Scott Kupor on the intrinsic-value fallacy, “price early, price often,” sales as the market sensor, and “the best place to sell something is where you have already sold something.” From the a16z Podcast, “Pricing, Pricing, Pricing.”

4. Kyle Poyar, Growth Unhinged: the abstraction layer, credits as a growth engine, and the platform-plus-credits hybrid model.

5. Gibson Biddle: the DHM product-strategy model (Delight, Hard-to-copy, Margin-enhancing).

6. Van Westendorp: the Price Sensitivity Meter.

7. Valueships x SaaSiest, State of UK AI SaaS Pricing (2025), and the SaaSiest Northern EU Benchmark Report (2025): the pricing-metric evolution, gross-margin compression across the spectrum, and the AI adoption gap (roughly 92% using AI, about 5% extensively, near half reporting no measurable impact).

8. Packaging strategies, feature allocation by adoption versus willingness to pay, the business-class analogy and the eight-step monetization process, from The Principles of Pricing (Notion Capital).

9. Commercieel Verbeteren, AI Pricing Playbook (Johan Maessen): the AI Value Maturity Framework and the overall synthesis.

10. Ames True Value Hardware & Supply, Route 1, Wiscasset, Maine, and the Maine mottos “Vacationland” and “The way life should be.”

Page updated

Google Sites

Report abuse