Stopped the losses, then earned the budget to scale

Situation

A Fortune 500 insurance brand selling auto and home insurance to consumers across the United States. This one is from inside the building: a senior in-house role I held, included here because it is the clearest example of the rebuild approach.

The Google Ads program ran on Search and Performance Max, with a large share of the budget sitting in Performance Max on a set-it-and-forget-it basis. When the search program lost its manager in late 2025, I picked up cover during the transition. A first pass through the account showed a program that needed a full rebuild rather than a caretaker, and the cover became the work.

What was broken

On the surface the account looked fine. It was running, it was spending, and it was losing money on that spend, week after week. The damage was not one big mistake. It was a stack of small ones.

• UTM parameters inconsistent across ads. The company’s own reporting could not cleanly tie a paid session back to the campaign that drove it.
• Keywords sitting in the wrong campaigns. An auto campaign could be paying to show up on a home insurance search, against the wrong message and the wrong landing page.
• Thin keyword coverage and no match-type strategy. The account missed real demand, and broad, phrase, and exact were applied with no plan for how they should work together.
• Headlines and descriptions not fully populated on a large share of ads, capping Quality Score and click-through rate.
• Performance Max running with no brand exclusion and no dedicated brand campaign. It absorbed branded searches that would have converted anyway and reported them as its own wins.
• Conversion tracking spread across too many signals. The account was not consistently optimizing toward the events the company’s martech team had approved.

The work

The rebuild ran on a reduced budget for a month to restructure the account before relaunching over a major holiday weekend. Every campaign was rebuilt from the ground up on a consistent naming convention. Conversion tracking was narrowed to the martech team’s approved events, UTMs were normalized, ad copy was fully populated, and keywords were moved into the campaigns that matched them.

The new structure was built around the account’s real drivers. A small set of high-volume head terms, searches like “auto insurance” and “home insurance,” carried a disproportionate share of conversions. Each major product was split into a core campaign and a dedicated exact-match campaign for those head terms, so the highest-value searches could be funded and watched on their own. A competitor campaign and a general insurance campaign were added, and the account grew from a handful of campaigns into a structure several times larger. Performance Max finally had brand separated from non-brand.

Before

Sprawling

• Much of the budget in set-and-forget Performance Max
• Keywords sitting in the wrong campaigns
• Brand and non-brand blended together

→

After

Segmented

• Core plus exact-match head terms for each product
• Competitor and general insurance campaigns added
• Brand separated from non-brand in Performance Max

The least obvious decision was splitting the match types at all. Google’s own team pushed back, since the standard advice is to consolidate and lean on broad match. In this category a handful of exact-match head terms are large and dependable enough to deserve their own campaigns and their own budgets. Kept separate, their impression share, click share, and spend could be managed directly instead of averaged into everything else.

Google’s own team pushed back on splitting the match types. In this category, a few exact-match head terms are too big to bury in broad match.

The result

The relaunched account performed from the first weeks. A program that had been losing money on every week of spend turned profitable.

That changed the question. A profitable account is one a company wants to feed, and this one did. Spend grew substantially from its pre-rebuild level, and the program held its footing as it scaled. It went from losing money every week to absorbing far more budget without giving the profitability back.

The turnaround was not one move. The restructure around the high-value head terms, the match-type split, and the conversion-tracking cleanup all landed in the same rebuild, and all of them were needed. The clearest evidence it worked is the budget itself. A program does not get funded to scale unless the people paying for it trust what comes back.

A program that was losing money every week became one the business chose to fund and scale.

What’s still true

The work is ongoing. The search program still runs on the structure built during the rebuild, with the core and exact-match campaigns anchoring the account, and there is still efficiency left to find.

The reporting stayed too. A daily paid media summary became the performance team’s working report, and parts of it, the daily tracker in particular, get used well outside the original program.