Nobody Said Take My Money — Part 7 of 7

Everything this series describes — the n=1 extrapolation, the year of unwitnessed building, the three demand tests that all came back “enthusiasm without purchase” — was learnable for under $500 and about a month of conversations. I paid fourteen months and a working product line instead. This final post is the whole failure inverted into the sequence I’ll run next time, written so you can run it Monday morning.

The spine of it is two rules. Everything else is application.

Rule one: commitment over enthusiasm. Every validation step must end in a binding act — a card on file, a paid beta, a signed pre-order. Politeness is not data. Enthusiasm is what politeness looks like at scale.

Rule two: sequence discovery from cheap-and-high-fidelity to expensive-and-low-fidelity. Conversations first, broadcast last. Paid ads amplify a message that already converts; they are a terrible instrument for discovering whether one exists.

The sequence

Step 1 — Ten strangers, past behavior only. Before building anything beyond a demo, find five to ten strangers with the problem. Not friends (Part 2: a friend is a doubly biased sample — loyal, and an outlier by selection). In the conversations, ask only about past behavior, never future intent: “tell me about the last time a predator got in — what did you do, what did it cost you,” never “would you buy this?” Future-intent questions harvest courtesy. Past-behavior questions harvest facts, including the fact that killed my product thesis (Part 5): the real losses mostly happened in ways my product didn’t address, and the audience already self-solved the rest with a free nightly walk.

End every call with a real ask — reserve a unit, join a paid beta, card on file. A 25-minute call that ends in “keep me posted” is a warm nothing. The phone’s failure mode is the opposite of the forum’s: forums are unaccountable, calls are too polite. The binding ask at the end is what squeezes truth out of both.

The test is pre-registered: five to ten binding commitments, or the extrapolation from “problem exists” to “market exists” fails, and the build doesn’t start.

Step 2 — Hand-sell to warm communities. If strangers who trust you (a forum where you have standing, a festival crowd, a farmers market) won’t buy from your hands, cold strangers won’t buy from your ads. The booth version (Part 4): a founder’s pre-order with a real card charged at ship, cancel anytime by text. Friction calibrated on purpose — not painless (a real card, a real form), but costless today. Ten cards in a weekend is a business; two is your answer. Add one question to the form — “what worried you enough to reserve?” — and the yes/no test becomes a yes/no-plus-direction test for free.

Step 3 — Channel before broadcast. The single best-evidenced item on this list. Months of broadcasting moved my venture less than one conversation with a prominent coop builder who wanted to include the product with his coops. Vendors already own the buyer’s trust and the moment of purchase intent; you’d otherwise pay to manufacture both. Target 15–20 premium coop builders (in your domain: whoever sells the expensive thing your product attaches to), by voice, pitched as differentiation — “smart monitoring, included, sells your $5k coop against the builder who doesn’t have it” — not as a margin split. Structure it so the vendor moves hardware and the subscription stays direct with you, with the vendor earning a recurring referral cut — then they’re paid to keep the customer subscribed, and you never lose the renewal.

And note the dual use: vendor outreach is itself a demand test. If builders won’t carry it when it costs them nothing and differentiates their flagship, that’s as loud as a failed campaign — obtained from fifteen phone calls instead of an ad budget.

Step 4 — Organic message tests. Post the pitch angles in the communities where you have standing and watch which framing gets pulled on. Free, and it locates the message before you pay to amplify one.

Step 5 — Paid ads, last. Only to scale a message that steps 2–4 already proved, and even then with the kill number written first. (A small, capped, pre-registered message probe against cold traffic is legitimate early — that’s a different tool from ads-as-growth-channel, and confusing the two is how I burned an agency engagement learning what step 1 would have told me for free.)

The standing orders

Write kill criteria before every test. “If the campaign misses X, development halts that day.” “If fewer than five of ten calls end in a commitment, no build.” Decided cold, in writing, before you’re invested in the result — because after the result, sunk cost will move any goalpost that isn’t nailed down. I had a clear, expensive, honest halt signal and built through it for six more months (Part 4). The number decides, not your feelings about the number.

Charge the beta users. My betas were free forever, and their passivity read as satisfaction when it was mostly just freeness — a free user costs nothing to keep and so produces no signal. Next time: hardware free (they’re doing QA), but the subscription — the number the business actually needed validated — charged from day one at a founder’s rate, roughly half price for life, in exchange for a written feedback cadence. Not a token dollar; a token dollar is a rounding error nobody bothers to cancel, so it tests nothing. Enough to make staying a decision.

Units shipped is the wrong metric; retained subscriptions are the business. The economics that took me a year to see: hardware sold at or near cost contributes zero. Two thousand kits shipped means nothing; two thousand retained subscribers at ~$30/month is roughly $220–300k a year of seller’s earnings and, at standard small-business multiples, a seven-figure exit. Design the recurring layer first; hardware is a customer-acquisition cost with a circuit board in it.

Razor-and-blades is right, and it’s leverage. I got to cost-hardware-plus-subscription eventually, but late, and without the guardrails that make it safe. (a) Lock the blade: the subscription must be structurally required for the core value — mine was, since the value was the remote alert, but verify this before subsidizing any future hardware, because a locally useful device with an optional cloud sub sells razors to people who skip the blades. (b) Near-cost hardware is a CAC, not a sale: giving hardware near cost to a customer who churns puts you underwater, so the model raises the stakes on retention validation rather than lowering them. (c) Never below cost until retention is proven, and offer annual prepay at a discount to pull cash forward and filter for committed buyers. My breakeven was 48 subscribers — the model was never the problem. The missing ingredient was anyone at all.

Treat taking-it-away as a demand instrument. The cheapest, most overlooked test in this series (Part 6): when something is removed — a feature deprecated, a pilot ended, a product sunset — the reaction is the purest demand reading you will ever get, because loss aversion is the strongest force in a buyer’s psychology. Silence at the removal is the same datum as silence at the wallet. If nobody fights the take-away, nobody wanted the thing.

The close

The friend was never the mistake. A real person with a real problem is the best origin a product can have — Mother Hen’s origin was sound, its engineering was sound, and its shutdown decision was sound. Everything broken lived in the untested leap between one man’s problem and a market’s demand, and that leap can be tested in a month, by anyone, for the price of some phone calls and a clipboard.

So: the friend is the hypothesis. Ten committed strangers are the test. Run the test before the build, write the kill numbers before the test, take commitment as the only currency of truth — and listen, at every stage, for the one sentence that never got said to me.

If you’re a year in and nobody has said “take my money” — nobody impatient, nobody pre-ordering, nobody trying to press a deposit into your hands before you’re ready — you don’t have a marketing problem. You have your answer. I hope this series makes it a $500 answer instead of a fourteen-month one.

— Tennis Smith, Smith Varmint Works

This concludes Nobody Said Take My Money. The platform sleeps intact; the lessons don’t.