Issue #1 — The AI that ate the ad budget

Google's own automated bidding was eating 20% of their monthly ad spend. They had no idea.

The account looked healthy from the outside. Click-through rate trending up. Impressions climbing. Quality scores green. Their previous agency's monthly deck was a parade of good-looking numbers. The only metric not moving was revenue.

The setup

A North American B2B industrial equipment manufacturer. Long sales cycles, real deal sizes, the kind of business where one new account moves the quarter. They came to us because they'd noticed something nobody on their agency's monthly call would explain: more clicks every month, more reach every month, same number of qualified leads in the pipeline.

The agency had them on a "trust the algorithm" setup — fully automated bidding, broad-match keywords, "let Google find your audience."

When we said we'd start with a search terms audit, the founder paused. "What's a search terms report?"

The find

The search terms report is one of the most useful screens in Google Ads. It shows you what people actually typed in to trigger your ad — not what you were targeting, what actually fired. Their previous agency had never opened it.

We opened it. The mess was loud.

Their broad-match campaigns were spending money on queries like:

• "[product] DIY"
• "[product] used"
• "what is [product]"
• "[competitor product] reviews"
• "how to repair [product]"

These are not buying searches. These are curious-people searches, hobbyist searches, problem-research searches. Nobody typing "how to repair [product]" is about to write a purchase order for a new one.

And here's the part that hurts: Google's automated bidding was optimizing for these. Why? Because the goal Google had been pointed at was clicks. Broad, low-intent searches click great. The AI was nailing its job — it just wasn't the right job.

The previous agency's monthly report was showing click-through rate up, impressions up, quality scores trending green. Everything was working — for the metric Google was watching. Meanwhile a fifth of their monthly budget was being lit on fire.

The fix

Three moves in week one:

1. Killed every broad-match campaign. Replaced with phrase- and exact-match keyword groups built around commercial intent. Queries like "[product] price," "[product] supplier," "buy [product]" — the kind people search when they're ready to issue a PO, not when they're killing time on a Wednesday afternoon.

2. Pulled automated bidding off the highest-intent campaigns. Manual bidding until enough conversion data exists to point Google at revenue, not clicks. The algorithm works — but only when it's optimizing for the right goal. We weren't ready to hand it over.

3. Rebuilt the account structure around the actual funnel. Awareness campaigns separated from intent campaigns. Different budgets. Different goals. Different definitions of success.

The result

The 20% of budget we recovered didn't vanish. We poured every recovered dollar into the commercial-intent campaigns that were already converting.

Within thirty days the account was running at 12.48x ROAS. Lead volume climbed to 131 new leads per month. The new monthly reporting line that mattered: zero spend on broad-match targeting.

Google's own AI had been the leak. Once we stopped paying it to find traffic, we started finding revenue.

The takeaway

If your agency talks about "trusting the algorithm" in monthly reviews, try one question on your next call:

"Show me last month's search terms report. What percentage of the queries we paid for had commercial intent?"

If they can't answer in thirty seconds, you have your problem.

Automated bidding works — but only when it's pointed at the right goal. Most accounts we audit are accidentally still pointed at clicks. The fix is rarely about spending more. It's almost always about spending less on the wrong searches.

---

Want us to look at your account? Book a free 20-minute audit. We'll find your version of this in twenty minutes — or you keep our full teardown playbook on the house.