Where 30 days of Google Ads spend actually went: 662 search terms, one small budget

We run a small Google Search campaign for Plyto. For 30 days ending September 13, 2026, we pulled every search term it showed for and sorted the spend by what the person was actually looking for. These are the numbers, with the terms named, so you can compare them with your own account.

The shape of the month

• 662 different search terms showed our ads
• 76 of them got a click; the other 586 showed and were ignored
• 101 clicks, about $320 of spend, an average of $3.15 a click
• 8 signups, so about $40 each

Every keyword in the campaign was phrase match. Google's close-variant matching decided which real searches those keywords showed for, and it decided broadly.

Where the money went

We sorted every term by intent. The percentages are share of spend.

• 26 percent: another product's name. Searches like "is bitrix24 really free", "hubspot pricing", "opensource mailchimp", "zeely alternatives free" and "scalio app". Someone typing a competitor's name is usually looking for that competitor. Twenty-five clicks, two signups.
• 22 percent: wanted a flyer, a poster, a video or a jingle. "how to make ai flyers", "ai banner generator", "ai commercials on tv", "commercial ai video". Plyto makes and runs ads on Facebook, Instagram and Google. It does not make a flyer for a garage sale. Twenty-two clicks, three signups, and every one of those three asked for a video or a jingle, then left.
• 21 percent: in the market for what we sell. "facebook ads automation tool", "ad generator ai", "ai ad". Twenty-three clicks, three signups. This is the fifth of the budget that was doing the job.
• 19 percent: hunting for something free. "free crm", "free email marketing platforms", "free ai funnel builder", "free email lists for marketing". Seventeen clicks, no signups.
• 12 percent: a how-to or how-much question. "how to make ai ads", "how do you run ads on facebook", "how much does it cost to advertise on facebook". Fourteen clicks, no signups.

Put the last two together: 31 percent of spend went to people who were not going to buy anything from anyone that day.

A signup is not a customer

Eight signups sounds fine for $320. It is not, once you read which searches produced them. Three came from the flyer-and-video bucket, and each of those people asked the product for the one thing it does not make, then closed the tab. One came from "pippit ai", a competitor's name. Of the eight, three were people shopping for an ads tool, which is the only kind of signup that turns into a customer.

If we had judged the campaign on signups, the flyer terms would have looked like winners. Judged on who came back, they were the most expensive clicks in the account.

What this costs a business that checks monthly

Six hundred and sixty-two terms in thirty days is twenty-two a day. Nobody running a two-truck plumbing company reads twenty-two search terms a day, and a $10-a-day campaign is too small for an agency to read them either. So the do-it-yourself searches, the job hunters and the competitor names run for the whole month, and the invoice arrives with the lesson.

The fix is not a cleverer keyword list. It is the daily loop: read yesterday's terms, block the ones that cannot become customers, move the budget toward the ad group that produced a real lead, and write down what changed. Done by hand it is a chore that stops after the second week. Done by software it is the same routine every morning.

What to do with a report like this

• Block the whole family, not the term. One negative for "flyer" catches "how to make ai flyers", "ai flyer maker" and the next twenty variants. Phrase-match negatives for flyer, poster, banner, video, commercial, jingle and script would have saved about a fifth of this month's spend.
• Block the questions. "how to", "how do", "how much" and "what is" are research, not shopping. Twelve percent here.
• Decide about competitor names on purpose. They can convert (one of ours did), but they belong in their own ad group with a page written for someone switching, not mixed into the general groups where they eat a quarter of the budget.
• Judge terms on who came back, not on who signed up. Three of our eight signups were people we could never have helped.

The caveats

One account, one month, a few hundred dollars. The percentages will differ in your trade. What will not differ is the mechanism: phrase match plus close variants puts you on searches you never bid on, and the only defense is reading the terms often enough to matter.

If you want to see your own account's version of this, the free Google Ads audit reads 30 days of your search terms and prices the waste in dollars, with no account needed. And this daily loop is what Plyto's agent does on every campaign it runs: how it runs Google Ads.