A 164-campaign account, rebuilt to 49.

The account covered 17 regional markets for a production homebuilder. Over several years it had grown into 164 separate search campaigns, most of them added one at a time to handle whatever was in front of someone that week.

Match types had been split into their own campaigns. A single market might run separate Exact, Broad, and modified-broad campaigns for the same keywords. Individual communities had been spun up as standalone campaigns rather than living inside the market they belong to. Naming conventions had shifted over time, so two campaigns doing the same job could carry different names. Bid strategies were scattered across portfolios, including legacy strategies still named after an import job from years earlier.

This is how most large accounts end up when edits get made under deadline and nobody owns the overall structure. The cost does not show up right away. It shows up later, in performance and in the hours the account takes to manage.

Splitting match types and communities into their own campaigns was once standard practice. The platform has changed since then, and the split now costs the account in a few specific ways.

• Smart Bidding learns slower. Value-based and target-CPA bidding need pooled conversion signal. When the same intent is split across three match-type campaigns, the algorithm sees three thin data sets instead of one usable one. Learning periods stretch and reset more often.
• Match types no longer earn their own campaigns. Modified broad was retired by Google in 2021, and close-variant matching means Exact and Phrase already overlap heavily. Maintaining a campaign per match type preserves a distinction the platform stopped enforcing.
• Community sprawl fragments small budgets. A standalone campaign for one community competes with its own market for budget and pulls spend below the volume any single campaign needs to optimize.
• Inconsistent naming blocks automation. Scripts, automated reporting, and bulk edits all depend on predictable names. Drifted naming means every report needs hand-cleaning.
• Management time scales with campaign count. 164 campaigns is 164 things to budget, pace, QA, and explain. The overhead is real and it is recurring.

The new architecture collapses the account along the line that actually matters to a bidding algorithm and to a buyer: search intent. Each market runs a small, fixed set of campaigns, one per intent bucket. Match types and individual communities move down a level into ad groups, where they organize keywords without splitting the conversion data.

None of these moves are cosmetic. Each one maps to how the auction and Smart Bidding behave now. Pooling the signal shortens learning periods. Matching a bid strategy to each intent lets brand, nonbrand, and competitor carry the different targets they should. Consistent naming is what makes scripted reporting and bulk edits possible at all. And with fewer campaigns, budget lands where there is enough volume to optimize against.

The output was the full plan, detailed down to the campaign and ad group:

• A full old-to-new migration map for all 164 campaigns, with the target campaign, ad group, and keyword treatment for each.
• The target architecture across all 17 markets: 49 campaigns, 92 ad groups, one consistent naming convention, and a bid strategy aligned to each intent bucket.
• The measurement design: per-market return targets and a bidding test structured to isolate the effect of the restructure.

The work shown here is the proposal and its rationale. Numbers are structural, drawn from the actual account. No post-launch performance figures are claimed.