Google Ads accounts built for manual bidding are now working against the algorithms that run them. Single Keyword Ad Groups, device splits and isolated match types made sense when a human adjusted every bid. Smart Bidding does not work that way.
It feeds on conversion volume, data velocity and flexible signals. When campaigns are too fragmented, the algorithm is left guessing. Accounts that feel jumpy, never exit learning and cannot scale usually have a structure problem, not a bidding problem.
The 30-conversion floor nobody budgets for
At campaign level, Smart Bidding generally needs at least 30 conversions per month to read contextual signals such as device, location, time and intent. Below that, it is optimizing on noise.
Take 60 monthly conversions spread across 12 thin campaigns. That is 5 conversions each. A single sale can make the algorithm overbid, while a quiet day can choke traffic off. The system cannot tell a real trend from random variance.
Fragmentation also creates budget traps. Fixed daily budgets on micro-campaigns cap high-intent surges while money sits unused in nearby quiet segments.
What legacy structures break
Three bottlenecks appear when an account is over-segmented:
- RSA testing stalls: Responsive Search Ads need continuous impression and conversion data to test headlines. Thin ad groups cannot feed that.
- Semantic friction: Smart Bidding matches intent, not exact strings. Dense cross-negatives left over from SKAG days block high-converting long-tail queries.
- Operational drag: Managing hundreds of tiny ad groups burns hours on low-impact maintenance instead of offer design or creative testing.
None of these are fixed by adding keywords. They are fixed by consolidating campaigns that represent the same customer journey.
Where consolidation should stop
Consolidation is not flattening everything into one campaign. Some splits reflect real business logic:
- Business objectives: lead generation and direct ecommerce need different conversion goals and bidding strategies.
- Margin variance: products with different profit margins need separate Target ROAS or Target CPA targets.
- Strategic geography: regions requiring dedicated budgets, localization or distinct positioning should stay separate.
- Funnel intent: high-friction demo requests and low-friction trials should not share a campaign.
The filter is straightforward: if a split changes your strategy, budget or creative experience, keep it. If it only changes how identical journeys are labelled in reports, merge it.
Make signal quality the KPI
Platform automation, including the expansion of features like AI Max, accelerates this shift. Broadening match, auto-generated assets and campaign-level automation remove the busywork manual segmentation once covered.
Start with an audit of SKAGs, duplicated match-type campaigns and low-volume segments. Then merge the redundant structures. Roll out the new structure on one product line or campaign group and benchmark stability and conversion volume before scaling.
The goal is not fewer campaigns for its own sake. It is giving machine learning enough dense signal to find high-value auctions.
Source: Search Engine Journal
