How much budget you need to test a paid channel

The budget you need to test a paid channel is about 30 conversions times what a conversion is likely to cost. That's the calculation, and it runs backwards from conversions rather than forwards from what you can spare. At a $60 cost per acquisition, that's $1,800. At $300 it's $9,000, and if that number scares you, you've learned something real before spending a cent.

Most channel tests die of something smaller than bad creative. A team puts $500 into Google Ads, gets four leads at $125 each, calls the channel expensive, and moves on. Four leads is noise. Run the same account three more weeks and it might settle at $70. That test didn't fail. It never ran.

How much budget to test a paid channel, in one calculation

Multiply the conversions you need by what each one is likely to cost. That's the whole formula. The work sits in the two inputs.

Conversions you need. Thirty is the working floor for a directional read. Below that, one lucky week swings your cost per acquisition by half and you'll read the swing as a signal.

What a conversion will cost. You've never run the channel, so you're estimating. Take the cost per click the platform quotes for your keywords and divide by the conversion rate you already see on the page you'll send traffic to. A $4 click and a 3% page means roughly $133 a lead. Multiply by 30 and the test is about $4,000. Use your own page's rate: the benchmark in someone's blog post was measured on someone else's traffic.

Then check that against what you can afford to pay. If your economics support $150 a lead and the estimate says $133, run it. If they support $40, the real question is whether you can cut the cost by two thirds, and you should know you're asking that before you start.

Google and Meta publish their own minimums

You don't have to guess at the thresholds. Both platforms say what their systems need.

Google's Target CPA documentation puts it plainly: "For evaluation, we recommend you measure performance for the last 30 days, including at least 30 conversions." That's Google telling you a month with 12 conversions isn't a verdict.

Meta asks for more, and the detail catches people. Its delivery system wants around 50 optimization events per ad set per week to leave the learning phase. Miss it and the ad set sits in Learning Limited, where results stay volatile and the numbers you're reading are mostly variance. The trap is "per ad set". Split a test across four ad sets and you've quietly asked for 200 events a week. In a first test, run fewer ad sets than feels right.

When the real conversion is too rare, test a proxy

Some businesses can't hit 30 purchases in a test at any budget. If you close eight contracts a year at $40k each, no ad spend produces 30 closed deals in a month, and it shouldn't.

So move the event you count up the funnel. Demo booked, trial started, qualified form submitted: pick the shallowest event that still tracks revenue, and check that it does before you trust it. Pull last year's demos and see what share turned into money. If a third of them close, demos are a fair proxy. If 2% close, you're about to spend a month buying demos that mean nothing, and the algorithm will get very good at finding more of them.

Write the kill criteria before you spend

Decide the rules before the first impression, in writing, in one sentence a colleague could hold you to. Without it, the call gets made on a Friday by whoever is most tired of looking at the spreadsheet.

We'll spend $4,000 over 30 days to buy 30 leads. Under $100 a lead we scale. Between $100 and $150 we fix the landing page and rerun. Over $150 we stop.

That sentence is the test. Everything else is execution. It also protects the channel from a bad week: you agreed to buy 30 conversions, so you don't get to quit at nine of them because Tuesday looked ugly.

What the test tells you, and what it can't

A clean test buys you one thing: a believable cost per acquisition. That's worth the money. It still won't tell you the channel is profitable, because profit shows up after the click, in your close rate, your contract size, and how long the cash takes to come back. We've written about how to work out that payback window, and it decides whether a $133 lead is cheap or ruinous.

The page decides as much of your cost per acquisition as the auction does. The same click at the same price converts at 2% or 6% depending on what the visitor lands on, which is why the landing page carries so much of the work in paid. Test a channel on a page you already know underperforms and you've tested the page.

So the sequence is short. Estimate the cost per conversion, multiply by 30, check you can stomach the number, write the kill rule, then spend the whole budget before you decide anything.

If you want a second pair of eyes on the math before you commit the budget, that's the kind of work we do.