The pricing page sat unfinished on my screen for a week.
Three economic models were in front of me. The Notion shape — freemium with tiers, a free tier, a $10 personal, a $20 team, a $30 enterprise. The Slack shape — per-seat SaaS, fine for teams and meaningless for solo users. The Costco shape — one annual fee, members get a differentiated experience, anyone can walk in. I drew each one out on paper, modeled the revenue against five thousand members at year two, and walked away from the desk because the question wasn't a math question.
Tiered freemium would have maximized revenue per customer. It would have also fragmented the brand into first-class and second-class members the moment the second tier shipped. Per-seat SaaS doesn't apply to a product designed around solo users. The one model that left the brand intact was the boring one — flat fee, binary access, no ladder.
$99 a year. One fee. No tiers, no Pro, no upgrade path.
The number is the same as a Costco Gold Star membership, rounded up by four dollars. Not by accident. The mental model I want a new visitor to reach for is the one they already know: "I pay Costco $65 a year and they curate the shelf for me." Microapp asks the same question in a different aisle. The brand goal is the membership card people would actually carry — the wallet hardware that signals belonging without performance. The price has to be the kind of number a member would tell a friend over coffee without needing to round it down to sound reasonable. Ninety-nine dollars passes that test. A hundred and ninety-nine doesn't.
The transferable why: a tier system optimizes revenue per customer but fragments the brand. A flat fee under-monetizes some customers but unifies the brand. Pick which problem you'd rather have for the next ten years. Most founders pick "fragmentation" by accident because revenue is the easier number to defend in a planning meeting. The other choice — flat, defensible, recitable in one sentence — is the one that compounds into trust.
Locked 2026-05-12 · $99/yr annual membership · live on /membership
The second decision was the hardest because the honest answer was uncomfortable. AI is not free to run.
A calculator microapp costs Microapp nothing per use — it's static JS that runs in the user's browser. An AI microapp can cost anywhere from a fraction of a cent to several dollars per run, depending on which model it routes to, how big the input is, and how thinky the answer needs to be. The cost spread inside the same product line is roughly 50x. Pretending that's a flat-rate problem is a lie that catches up with you the day a popular tool burns through the month's compute budget.
The standard SaaS dodge is "unlimited AI" with a quiet rate limit. The standard token-billing dodge is metering after the fact — the user runs the tool, gets a bill at the end of the month, learns they spent $14 they didn't plan to spend. Both options dump the uncertainty onto the customer in different ways. The version that respects the member is the one that puts the cost on the page before the run.
AI runs on credits. Every AI tool publishes its price. Members see the number before they hit the button.
Members pay $99 a year and get a generous credit allowance every month — enough that most members never see the meter. When the allowance runs out, top-ups are available at near-cost, members only, never the consumer-AI margin. Non-members get a one-time taste — a small allowance per device, a larger one with a free account — and then membership is the only path past the taste. Non-AI tools stay free forever for everyone.
The transferable why: when your costs are variable, your pricing has to be too — but the variability has to be visible to the customer before they incur it. Flat pricing on variable-cost services is a leak; meter pricing without pre-disclosure is a surprise. Pre-disclosed metered pricing is the version that builds trust. The customer who knew the price before they paid it doesn't feel cheated by it.
Locked 2026-05-12 · credit pricing visible on every AI tool's page
The third decision was the one I was most tempted to walk back. The binary.
Every pricing meeting I've ever heard of ends with someone proposing a third path. The $4.99/mo entry. The $19.99/mo Pro. The à-la-carte credit pack for non-members who don't want to commit. Each one is defensible in isolation. Together they turn the brand into a ladder of progressively bigger versions of itself, with the corresponding committee of marketing variants that have to be maintained.
Costco doesn't have a Costco Pro. Costco doesn't have a monthly option. Costco has Members and not. The friction of joining is exactly the part Costco refuses to soften — because the friction is the brand promise. "You decided to join. In exchange, you get the curated shelf."
You're a Member or you're not. No third path.
No à-la-carte credits for non-members. No monthly membership option until enough members specifically request one. No Pro tier ever — same membership for everyone, indefinitely. The simplification is the point: one decision, two outcomes, recitable in one breath. "Members get clean pages and AI at cost. Non-members get the same tools with ads." That sentence is the entire pricing page.
The transferable why: every ladder rung you add to your pricing is a place the brand has to argue with itself. Brands that survive a decade tend to be the ones where the choice is binary — Members or not, Owner or not, Subscriber or not. Ladders feel like they widen the market; they more often widen the engineering and marketing surface that the market has to navigate before deciding. Pick binary, and the rest of the company gets simpler in the same direction.
Locked 2026-05-12 · no third path · enforced by the /membership covenant
The fourth decision is where the model genuinely differs from SaaS, and I almost missed it. It's about which side of the paywall the tool itself sits on.
SaaS puts the paywall between you and the tool. Cancel a Notion subscription and your databases go read-only. Cancel an Adobe subscription and Photoshop opens to a paywall screen. The product you used is held hostage to the next payment. The pattern maximizes retention via friction — you stay subscribed because leaving is expensive in a way that has nothing to do with the value the product provides.
The Costco paywall sits in a different place. You can walk into Costco's parking lot. You can buy from the food court without a card. The card unlocks the warehouse — the curated, differentiated experience — not the existence of the building. Cancel your membership and Costco is still there. Different aisles, different prices, but still there.
The paywall sits between members and the experience, never between users and the tool.
Every microapp opens for everyone. Non-AI tools are free, with ads, forever. AI tools give a taste before asking. Cancel a Microapp membership and every tool you used is still there at the same URL, working the same way, just with ads back on and no AI credits. Nothing is held hostage to the next $99.
The transferable why: where you put the paywall is a brand statement, not a pricing decision. Paywall between user and tool = SaaS. Paywall between member and experience = Costco. The first optimizes for churn rate this quarter; the second optimizes for the long-running brand reputation that compounds into word-of-mouth. The two strategies look similar on a P&L for the first year and diverge violently after that.
Locked 2026-05-12 · open-on-cancel · no tool is gated behind membership
The last decision is the one that turns the model from a pricing page into a brand. The ten percent pledge.
Most companies that "give back" give a percentage of profit, and the profit is whatever the accountants say it is. Patagonia gives 1% of sales — gross revenue, not profit, not adjusted, not "after we've paid ourselves." Microapp adopts the same shape and triples the number. Ten percent of revenue, off the top, audited quarterly, recipient causes chosen by member vote once the pool is large enough to be worth voting on.
$9.90 of every $99 membership leaves the company before anything else.
The math is unambiguous on purpose. Members can compute the giving number from the published membership count. Skeptics can audit the quarterly report against the published math. The pledge is the brand statement that's hardest to fake — and the brand statement that's most expensive to walk back once said.
The transferable why: a giving claim measured against profit is measured against a number the company controls. A giving claim measured against revenue is measured against a number the company doesn't. The version you can't fudge is the version members can trust — and the version you'll be forced to keep when the accountants would rather you didn't. Pick the version you can't fudge, and the pledge does the marketing for you.
Locked 2026-05-06 · BRAND.md §7 · quarterly audit published on /pledge
That's the model. One fee, real numbers, no third path, no hostage tools, ten percent off the top. Every other decision in the operating chapters runs against this skeleton.