You turn on your first Google Ads campaign the week after launch, because someone told you paid search is the fastest way to learn whether people want what you built. Two weeks later you have spent your test budget, converted almost nobody, and you're deciding whether the whole channel is broken or whether you did it wrong. Neither question has an answer yet, because you were asking the campaign to do a job it can't do this early.
The first campaign's job is not revenue.
A brand-new account has no conversion history, no audience signals, and no sense of which query actually describes your product. Google's own bidding systems learn from data you don't have yet. Judging a first campaign on cost per acquisition is judging a thermometer for taste, and most founders who conclude "paid search doesn't work for us" reached that verdict from a test that was never built to prove it either way.
What a first campaign can do is tell you what your product is actually called, in the words a stranger would type. Founders write ad copy in the language of their pitch deck. Searchers type in the language of their problem. The gap between those two vocabularies is the single most useful thing a month of paid search can teach you, and it shows up nowhere else, not in customer interviews, where people are polite, and not in your own analytics, which only sees people who already found you.
Budget it as research, not as a bet on return.
Set the number you're willing to spend before you look at any dashboard, and treat it the way you'd treat a research budget: gone at the end, on purpose, in exchange for information. This reframing changes what "disappointing" means. A month that produces three customers and a clear list of which phrases converted and which didn't is not a failure. It's the research working. A month that produces zero customers and no clarity on language is the one to worry about, and it usually means the ad groups were too broad to teach you anything.
Keep the test small and the vocabulary list wide. Five or six ad groups, each built around one way a founder might describe the problem you solve, will surface more than one broad campaign optimized for volume. You are not trying to win an auction yet. You are trying to find out which auction you're even in, and a narrow test answers that faster than a wide one, even at the same spend.
Negative keywords do more work than the campaign itself.
Before you have positive signal, the fastest way to waste a test budget is showing up for searches that were never yours. A startup selling project management software will burn through a week's spend on people looking for a job title, a movie, or a much bigger competitor's brand name, none of which was the plan. Building the negative list early is the difference between a test that teaches you something and one that just teaches you Google Ads is expensive.
Check search terms daily for the first two weeks, not weekly. Every irrelevant query that slips through at this stage compounds, because the algorithm reads the click as a signal to show you more of the same, and a week of drift is harder to undo than a week of restraint would have cost you.
Read the account for questions, not for a verdict.
At the end of a first month, the right output is a short list: which phrases produced a click that looked like your actual customer, which produced nothing worth learning from, and which surprised you. That list is worth more than the campaign's raw numbers, because it tells you what to test next and what to leave alone.
A startup that treats a first Google Ads campaign as research ends month one with a vocabulary list and a shorter negative-keyword problem. A startup that treats it as a revenue channel ends month one deciding paid search doesn't work, based on a test that was never built to prove that either way. The account looks the same either way. What differs is whether anyone learned from it.
We'd start the same way with any early Demand engagement: a small, deliberately narrow test, read for language before it's read for return. Tell us what you're building.