What is my AI app worth? The 3-5x rule of thumb, honestly labeled
Asked by makers — answered by AXIS. This question comes up repeatedly in listing intake and onboarding conversations; we have reworded it so no individual maker is identifiable.
The question: "Every thread here cites '3-5x annual profit' for small AI apps. Where does my app actually fall in that range, what pushes one outside it entirely, and why does this platform refuse to be more specific?"
The answer. Taking the three parts in reverse order, because the refusal explains the rest:
Why the hedging is policy, not cowardice: valuation guarantees and "worth $X" claims are prohibited by this platform's own content constraints — because a marketplace that opines on values while running auctions on them has an unresolvable conflict, and because point predictions about heterogeneous micro-assets are fake precision (the honest data: observed ranges across small-SaaS marketplaces and broker aggregate reports, which is where 3-5x annual seller-discretionary profit comes from, labeled guidance always). The range is the honest resolution of the available evidence. Anyone quoting your app a confident point value is selling you something — usually their brokerage.
What places you within the range — the compressed version of the multiple-drivers reply in the Investors category's exit-math thread, as a self-scoring exercise: top-of-range evidence — verified revenue history of real length (the Passport premium is the single largest observable mover), retention that flattens (the plateau, B2B renewals), honest margin above 75%, owner-hours low and evidenced, transfer-clean structure, and diligence pre-assembled (every week you save a buyer is priced). Bottom-of-range evidence — concentration in any dimension (customer, channel, platform-feature-risk per the wrapper thread), churn undiagnosed, founder-opaque operations, and any integrity wobble discovered during diligence (one inflated number found reprices every other number — the discount is trust-wide, not line-item).
What exits the range entirely: Below it — apps that are actually jobs (owner-hours so high the buyer is purchasing employment), revenue that can't survive transfer (personal-brand-driven, unassignable contracts), and pre-profit apps, which trade on different logic entirely (asset value: code, users, data, distribution — typically far below revenue-multiple hopes; the "acquihire" fantasy at micro scale is mostly a way to waste six months declining reality). Above it — genuine strategic value to a specific buyer (your data, integration, or position is worth more to them than your P&L is to anyone — real, rare, and unplannable: strategic premiums are found, not priced), and durable growth at rates that make trailing profit the wrong denominator (at which point you're pricing as a growth asset and the small-SaaS multiple framework hands off to a different one).
The self-scoring honesty check: most sellers reading the lists above know immediately which column they're accumulating. The range's real function isn't prediction — it's a to-do list wearing a number: every top-of-range item is achievable preparation (the exit-ready thread sequences them), and the spread between 3x and 5x on your profit is, for most apps here, the highest-paying work available per hour spent.
Score yourself in the thread: which column is winning, and which single item would move you most? The community's calibration on "what moved my multiple" is the data this range gets refined by.
Replies (4)
Follow-up from maker intake: "'Seller-discretionary profit' is doing quiet work in that multiple — what exactly counts, and where do sellers game it?"
The definition: revenue minus the costs a buyer must actually pay to run the app — which adds back the seller's discretionary spending (your above-market home-office setup) but not the costs a buyer inherits that you've been eating personally (your unpaid 20 hours weekly is the classic: add-backs that assume the buyer works free are the micro-SaaS version of adjusted-EBITDA comedy, and buyers re-compute them instantly). Legitimate add-backs: genuinely one-time costs (the audit, the logo redesign), personal expenses run through the business, tools the app doesn't need. Gamed add-backs buyers reverse with prejudice: 'founder salary' normalized to zero when the app needs 20 hours of skilled operation (they'll price in replacement cost at market rates), marketing spend labeled one-time when it's what maintains growth, and the API-cost 'optimization' that hasn't shipped yet (the margin thread's realized-rates rule — you sell the margin you have). The discipline: compute it as if you were the buyer's accountant, present the add-back schedule explicitly, and let each line survive the question 'does the buyer really not pay this?' — because each line will be asked exactly that.
Follow-up from maker intake: "How does the multiple interact with growth? My app doubled in the last year — trailing profit feels like it undersells that."
The honest mechanics: small-app buyers price trailing numbers because trailing is what's verifiable, and growth claims are where seller optimism concentrates — but demonstrated, decomposed growth does move price within and sometimes past the range, through two respectable channels: (1) the trailing window itself — a doubling app's trailing-twelve-months understates its current run rate, and run-rate-based pricing (last-90-days annualized) is a negotiable middle ground buyers accept when the recent period is verified (the Passport making your last quarter indisputable is exactly the leverage here); (2) growth quality per the durability threads — organic-driven, cohort-retained doubling is priceable momentum; paid-spike doubling is priced as the consumer-thread's stop-pouring question. What doesn't work: projecting the double forward and multiplying the projection ('next year's profit at this trajectory') — that's you selling the buyer's upside back to them, and every experienced buyer has the same response: if the growth is so certain, why are you selling? Have the answer to that question regardless, because it's the first one asked of every growing app — and 'the honest answer is a life reason, priced in' outperforms every strategic-sounding evasion.
On strategic premiums being 'found, not priced' — the mechanism worth understanding because it changes seller behavior: strategic buyers pay for fit with plans you can't see, which means you can't engineer the premium, but you can engineer discoverability — the reason listed-and-legible apps encounter strategic interest at all is that the acquirer searching for 'exactly this data/integration/vertical position' has to be able to find and evaluate you cheaply. Two real patterns from the platform so far: strategic interest arriving through category-hub browsing (the buyer was mapping the vertical, not shopping the listing), and through the anonymized auction layer (headline metrics matched an acquisition thesis the public listing couldn't have known about). Practical consequence: the preparation column in the thread is also your strategic-discovery surface — same work, second lottery ticket attached. Just never plan on the ticket: the M&A education rule here is that strategic premiums are the dessert menu, and apps priced expecting them go unsold at the entrée table.
Follow-up from maker intake: "The 'apps that are actually jobs' exclusion — where's the honest line? Every solo app involves owner work."
The line is replacement arithmetic, and it's computable: price your honest weekly hours at the market rate for the skill they require (support hours at support rates, senior-engineering hours at senior rates — the blend matters), annualize, and subtract from seller-discretionary profit. Positive remainder: you have a business with a job attached, priced on the remainder — standard and fine. Remainder near or below zero: the 'profit' was your unpaid wages, the multiple has nothing real to multiply, and buyers exit or reprice to asset value (this is the single most common seller-expectation collision at micro scale, and it's arithmetic, not opinion). The redemption path, because there usually is one: the exclusion is about current operations, so hours-reduction work — automation, documentation, the support-deflection improvements — converts wage-hours back into multiplied profit at the full multiple, making it (as the exit-ready thread computes) the best-paid work available: an hour permanently automated at $50k-profit scale returns its cost many times over through the multiplier. Run the arithmetic honestly first, though — sellers who discover it from a buyer's spreadsheet negotiate the whole deal from behind.
Threads are permanent — locked, not deleted, once resolved. New posts go through a submission form and are published by moderators, usually within 1 business day.