Run one small paid ad test, then scale

Why this matters

Earlier steps in this checklist lean on free channels. A small paid test answers the one question free channels can't: what does a customer cost when you pay for attention? Skip it and you'll either scale ads blind later, or avoid paid forever and leave a working channel on the table.

What "done" looks like

• One channel picked — not three
• One offer, one audience, one landing page, written down before spending
• A fixed budget ($200–$500) spent over 2–4 weeks
• Tracking that ties each lead or sale to the test
• A written result: cost per customer vs. your gross margin

How to do it

1. Pick one channel where your buyers already look: Google Search for people hunting a fix, Meta/Instagram for discovery, a local sponsorship for foot traffic.
2. Write one offer with a single call to action.
3. Set a hard budget and end date — $200–$500 over 2–4 weeks teaches plenty at small scale.
4. Install tracking before spending: conversion tracking, a promo code, or a "how did you hear about us?" field.
5. Let it run for at least two weeks; daily fiddling ruins the data.
6. Compute cost per customer and compare it to margin. That number drives the "double down" step ahead.

Common mistakes

• Splitting $300 across three channels — nothing gets enough data to mean anything
• Judging by clicks and likes instead of cost per lead or sale
• Killing the test on day 3, or letting an obvious loser run for months

Real-world examples

• Dropbox measured early paid search at $233–$388 to acquire a customer for a $99/year product — that test told them to drop ads and build their referral program instead.
• Google Ads has no minimum spend, so a tightly capped daily budget is a real product feature, not a workaround.

From a founder's point of view

A paid test isn't a bet that ads will work — it's buying a number. Cost per customer is the most useful figure in a marketing budget. Even a test that proves a channel is too expensive is cheap, because it prevents a much bigger waste later.

Rule of thumb

If you can't say "one channel, one offer, $X spent, $Y per customer," you ran ads — you didn't run a test.