Cutting paid search spend when the board asks for it

A client’s board asked the company to bring paid search spend down and to justify what was left from the bottom up. The quickest lever in Google Ads is the daily budget on each campaign, so that’s where one of the first cuts went. Within a week, the conversion the bidding relied on had fallen by close to 90%, while daily spend barely changed.

In an account bidding toward a return target, budget caps are the wrong tool for spending less. Decide which keywords, ad groups and campaigns to stop buying, using closed deals as the test, and judge the cut on the rate from qualified lead to closed deal, since the lead count is supposed to fall. The review sheet I now run before any cut is at the end.

What the board asked for

For a few months, more paid search budget had been going into lower-quality traffic without matching results. Ahead of the next board meeting, the CEO set the brief. The board wanted the remaining paid spend justified from the bottom up, based on what each part of the account produced, and a lower monthly run rate reached in stages. The CEO wanted the board to see the work behind the number.

The first week went the way a cut should. The paid search manager paused high-spend keywords with no closed deals behind them, paused ad groups in the competitor campaign that weren’t producing closed deals, and tightened match types. That week non-brand spend fell 22% while qualified leads rose 40%, helped by seasonal lift. Performance Max spend came down by about 70%, and its qualified leads went up.

What lowering the budget caps did

Two weeks in, the month was pacing above target. Early in the third week, two non-brand campaigns, one of them the competitor campaign, spent at their usual rate and produced almost no qualified leads. The paid search manager did what looked reasonable and lowered those campaigns’ budget caps.

Over the next two days, impression share on the competitor campaign dropped sharply. Impression share lost to budget spiked even though daily spend was about the same as before. Google was reporting auctions the campaigns could have entered and didn’t. By the end of the week, the primary optimization conversion, a qualified lead of any value, had fallen by close to 90% week over week.

Google had recently changed how its bidding treats budget-limited campaigns. It now aims for the return target over a 30-day window, and a sudden budget constraint was variability the campaigns hadn’t seen in a while, so they entered fewer auctions. Someone on the call asked whether, with the next six weeks this important, we should be changing budgets at all. We had already told leadership to expect roughly half as many leads and a much smaller drop in closed deals, and leads from earlier weeks were still moving through the later sales stages at a good rate.

How we cut spend instead

The competitor campaign’s caps were raised the next week. New return targets and budgets went in and were left alone for two weeks so the bidding could relearn. At the end of that month the team agreed that the next month’s cuts would come from removing specific campaigns, ad groups and keywords, with budget caps left alone. The paid search manager pointed to the year before, when the account had been cut roughly in half with little loss of volume.

Keywords tied to closed deals

Google Ads could show which keywords produced qualified leads, but closed deals happen weeks later in the company’s own systems. Using a query I wrote, the data lead pulled a warehouse report tying each keyword and campaign to qualified leads and then to closed deals, with new and existing customers split out. That report changed two decisions. One competitor keyword with a good record of closed deals kept its presence, held at about 30% impression share with a slightly higher return target. Another competitor keyword turned out to be one of the most expensive in the account with the fewest closed deals behind it, so it was cut further. A test on a third competitor term produced one closed deal, with a deal size about 40 times what the test cost. One deal doesn’t make a pattern, so it stayed a test.

The conversion the bidding learns from

Earlier in the year, non-brand search had been set to optimize toward a later qualification step, the event closest to revenue. It happened too rarely, and its values varied too widely, for the bidding to learn from. Performance Max, optimizing toward the earlier qualified lead, was sending through leads who did better at that later step than the campaigns aimed at it. A few months before the cut, we moved search back to the qualified-lead goal and kept passing the later stages back as values. A few settings needed attention as well. The value assigned to a qualified lead had been set most of a year earlier and no longer matched close rates or deal sizes. One funnel stage was labeled as a purchase in the ad account. Disqualified leads were also being sent back to Google as negative signals, though the effect was hard to see.

Brand search

Cutting Performance Max helped brand. With less overlap, branded clicks rose about 50% over 30 days. Some brand conversions turned out to be returning customers, which flatters paid brand as a source of new business. A geographic holdout was worth running, but not while so much else in the account was changing. The next month the team paused the exact-match brand keywords, and I built a weekly brand tracker to watch whether organic search picked up those clicks. One unrelated company was bidding on the brand name, and the plan there was to ask them to stop.

Broad match and missing exclusions

Pausing a keyword didn’t always stop the ads it used to trigger. Paused keywords were still matching through broad match into unrelated auctions, including one theme with a very low closed rate. The fix was negatives across every similar theme, with broad match kept only where the core query produced qualified leads. Where all the traffic came through the exact term anyway, the match type was switched. Months earlier, a URL exclusion set up for AI Max had never been added to Performance Max. I now check that every exclusion exists in every campaign type that can spend money.

Staging the second month’s cut

The second month still ran hot. About half the month’s budget was spent in the first eight days, which is one cost of taking caps off the table. The paid search manager bid down on most keywords while keeping the strongest competitor term, and I sense-checked the new return targets before they went in. In the third week of that month, spend was down 37% week over week while total qualified leads rose, and non-brand plus Performance Max was already spending at the target run rate.

How we measured it and what each team saw

Each group watched a different number, and each was accurate.

The timing gap sits between the first two rows. About a quarter of this client’s customers closed within three days of signing up, half within a week and three quarters within 30 days. A seven-day read understates any change, so we read closed deals on a window of at least 14 days and expected the first half of the target month to look different from its second.

The data lead’s test comes down to one question. Is the rate from qualified lead to closed deal rising as volume falls? If it is, fewer leads is the plan working. If it isn’t, the cut removed good buyers along with the waste.

Results

The first month of the new approach was the first at or near spend-positive for paid search, measured as the value of new customers against spend, using an assumed customer lifetime value. The second month tracked about the same.

Two things aren’t settled. The rate from qualified lead to closed deal was still being tracked when I wrote this, so I can say paid search reached about break-even on that measure, but I can’t yet say every cut removed only waste. And the efficiency hasn’t turned into growth. No channel, paid or otherwise, has yet shown it can take the freed-up budget and grow new customers. Where that budget goes next is now the main question.

What I’d do differently

Build the closed-deal report before the first cut. The full version arrived at the end of the first month, after two rounds of changes. With it on day one, I think we’d have skipped the detour through budget caps.

Write the cutting rule down before anyone touches the account. In a value-bidding account the rule is to remove keywords, ad groups and campaigns and leave the budgets alone. It only became the written rule after a bad week.

Tell sales what’s coming, with numbers. Leadership knew to expect about half the leads, and sales still felt it as inbound drying up. A short note with the expected drop in leads, the closed deals that should hold and the date of the check would have presented the slowdown as part of the plan.

Change one thing at a time. Budget caps, new return targets, a new search channel and a brand pause all landed within a few weeks of each other. That overlap is why the brand holdout had to wait.

Review sheet to use before cutting paid search

I run this before any paid search cut now. Each row is a check, how to run it, and what result would change the plan. Download the template.

The reporting questions behind this case are covered in when marketing and sales report different lead numbers, why B2B lead generation produces leads but not revenue and how to report growth to a board.