Fix: Search campaigns over-split — fragmented Smart Bidding signal

Your audit flagged that this account runs an over-segmented Search structure — either Single Keyword Ad Groups (SKAGs) at scale, or more than ~20 campaigns covering near-identical themes. The pattern made sense in the 2014-2018 manual-CPC era, but with Smart Bidding now the default and Google evaluating performance on 30-conversion-per-30-day windows [1], the same structure starves auction-time ML of the signal it needs to optimize [2][3]. This article explains what counts as over-split in 2026, why the modern consensus points to consolidation, the narrow cases where additional splits still earn their keep, and how to merge without losing data.

Why this matters

The over-split pattern descends from a manual-CPC playbook that worked when humans set bids per keyword. SKAGs — one exact-match keyword per ad group, tightly themed copy, controllable bid — gave the manager precision over Quality Score, CTR, and CPC. That model assumed the manager was the optimizer. In 2026, the optimizer is Smart Bidding, which optimizes against auction-time signals that only become statistically usable once enough conversions land on the same campaign-level model [2][3].

Google's evaluation guidance is explicit on the data threshold: recommended measurement windows are at least 30 conversions, or 50 for Target ROAS, before drawing conclusions about a Smart Bidding strategy [1]. When 40 sibling campaigns each book 3-5 conversions per month, none reach the floor — and the ML never leaves the noise band. Practitioner write-ups from 2025 quantify the symptom: splitting traffic into hundreds of micro ad groups "provides the system with mere scraps of data, preventing it from ever gathering enough information to learn" [6].

The Hagakure framing formalises the consolidation argument. One documented case study reduced a shoe retailer from 555 campaigns to 3, achieving roughly 300% search growth on the same creative — explicitly attributed to consolidating data silos into Smart-Bidding-readable units [2]. Brad Geddes' Adalysis writing in late 2025 reinforces it: "your biggest protection is your campaign structure", and reducing asset combinations through deliberate consolidation lets machines learn faster than testing thousands of variations across fragmented containers [5].

Google clarified the official stance in February 2026. Brandon Ervin (Director of Product Management for Search Ads) framed it: "Consolidation is not necessarily the goal itself. This evolution we've gone through allows you to get equal or better performance with a lot less granularity" [4]. His shared reference point — 15 conversions across 30 days as data-sufficiency, aggregable across campaigns via shared budgets or portfolio strategies — is roughly half the textbook 30-conv floor but tells the same story: signal density matters more than container count.

Practitioner counter-arguments deserve airtime. Granularity still earns its keep when (a) two product lines have materially different conversion values and tROAS should weight them separately, (b) seasonality patterns diverge — Black-Friday-skewed lines should not share a campaign with evergreen lines, (c) geos differ by material budget allocation and need per-geo pacing, or (d) regulated industries where ad-copy compliance differs by product family. The right number of campaigns is the smallest set giving each campaign a lever no other campaign provides.

How to verify the issue

1. Open Campaigns, filter to Search type only, status = Enabled.
2. Sort by Conversions (last 30 days). Count how many enabled Search campaigns book fewer than 30 conversions in the window.
3. If that count exceeds half the Search campaigns in the account, the structure is signal-starved at the campaign level — the clearest over-split symptom.
4. Drill into ad-group structure: open one or two lower-converting campaigns and check the Ad groups tab. One keyword per ad group (or ≤3 keywords) across many ad groups is the SKAG pattern.
5. Look for theme duplication: compare campaign names and primary keywords. If two campaigns target near-identical themes (e.g., "Running Shoes - Men - NYC" and "Running Shoes - Men - National excl NYC") with the same bid strategy and similar budgets, ask whether the split is doing meaningful work.
6. Check bid strategy status in Tools → Shared library → Bid strategies. Portfolio tCPA / tROAS applied across many low-volume campaigns is a strong signal that further consolidation would help [4].
7. Practitioner benchmark, not Google-published: if your Search account has more than ~20 campaigns on monthly spend under $10k, the structure is almost certainly too granular for current Smart Bidding economics.

How to fix it

1. Map each campaign to a distinct business lever. For every Search campaign, write down which it uniquely controls: budget tier, geo with materially different CPL, language, conversion value tier (low/mid/high AOV), regulatory ad-copy requirement. If a campaign owns none of these, it is a merge candidate.
2. Group merge candidates by intent and conversion economics. Campaigns serving the same intent (brand defense, non-brand acquisition, retargeting) with similar values and CPL belong in one container — see Campaign structure logical for the broader principle.
3. Plan the merge target. Pick the strongest performer in each merge group as the receiver. Migrate keywords (consolidating SKAG ad groups into Single Theme Ad Groups holding 3-15 semantically related keywords each), then move ad assets and audience signals.
4. Pause sources, do not delete. Pause merged-out campaigns rather than removing them — you retain historical data and have a fallback if the consolidation underperforms.
5. Reset the bid strategy. If the receiver was on tCPA or tROAS, leave the target unchanged for the first 7 days. Smart Bidding re-enters its learning phase after a structural change, and Google recommends evaluating over at least 30 conversions or one month [1] — see Smart Bidding learning phase.
6. Hold the new structure untouched for 14 days. No budget swings >20%, no tCPA/tROAS edits, no new conversion actions. Interventions during the window mask whether the structure change worked.
7. Re-evaluate at day 30. Compare campaign-level CPA / ROAS / impression share against the pre-merge baseline. If the receiver now clears 30 conversions / 30 days and CPA/ROAS hold or improve, the consolidation worked. If CPA drifts upward on a specific theme, that theme genuinely needed its own container.

How to confirm the fix worked

• Every active Search campaign clears 30 conversions / 30 days at the campaign level [1].
• No two Search campaigns target near-identical themes with the same bid strategy and similar budgets.
• Each campaign owns at least one distinct business lever (budget, geo, language, value tier, regulatory copy).
• Ad groups inside merged campaigns hold 3-15 semantically related keywords each — no remaining SKAGs.
• Smart Bidding status in Campaigns → bid strategy column shows "Eligible" rather than "Learning" or "Limited by data" after the 14-day hold.
• Account-level conversion volume held or grew through the merge — confirming consolidation did not suppress demand capture.
• Re-running the Whitead audit clears search_campaigns_over_split.

This finding rarely sits alone. Accounts that fail search_campaigns_over_split almost always co-occur with smart-bidding-learning-phase lingering on several campaigns, bidding_strategy_goal_alignment mismatches (tCPA on volumes below threshold), and negative_keywords_missing made worse by fragmented negative-list management. Clearing the over-split typically quiets two or three downstream findings on the next audit because the remaining campaigns finally cross the data-sufficiency floor. Sequence the consolidation before tuning bid targets — moving structure first means the targets you set afterwards reflect the post-merge economics, not the noisy pre-merge averages.

The blast radius is moderate but real: any structural change resets the learning phase, and over-aggressive consolidation (collapsing genuinely different conversion economics into one container) trades fragmentation for averaging — neither is a win. The 30-day post-merge evaluation window is non-negotiable; judging at day 7 risks reading learning-phase noise as structural failure and reverting prematurely.

Related rules + concepts

• Smart Bidding — the underlying optimizer that benefits from signal aggregation and suffers from fragmentation.
• Smart Bidding learning phase — the post-merge stabilization window and how to read its signals.
• Campaign structure logical — the broader rule covering goal-aligned campaign structure; over-split is a specific failure mode under it.
• Brand vs non-brand separated — a legitimate reason to split that survives the consolidation pressure.
• Search query mining — once merged, query-level hygiene (matched search terms, negatives) becomes the next lever.

Sources

1. Google Ads Help — About Smart Bidding. https://support.google.com/google-ads/answer/7065882 (accessed 2026-05-27). Documents the 30-conversion (50 for tROAS) evaluation windows and confirms Smart Bidding can pool signal across campaigns.
2. Search Engine Land — The Hagakure method for Google Ads management (Benjamin Wenner). https://searchengineland.com/hagakure-method-google-ads-management-432867 (accessed 2026-05-27). The case study reducing 555 campaigns to 3 with documented growth attributed to consolidation.
3. Search Engine Land — When to restructure your Google Ads account — and how to do it right (Dii Pooler, 2025-08-14). https://searchengineland.com/when-to-restructure-your-google-ads-account-and-how-to-do-it-right-460698 (accessed 2026-05-27). Modern consolidation-first guidance and the fragmentation warning.
4. Search Engine Journal — Google Clarifies Its Stance On Campaign Consolidation (2026-02-12). https://www.searchenginejournal.com/google-clarifies-its-stance-on-campaign-consolidation/567295/ (accessed 2026-05-27). Quotes Brandon Ervin (Google) on consolidation as a means, not an end, plus the 15-conv / 30-day data-sufficiency threshold.
5. Adalysis — Managing Google Ads accounts in the AI era (Brad Geddes, 2025-10-22). https://adalysis.com/blog/managing-google-ads-accounts-ai/ (accessed 2026-05-27). Brad Geddes on structure as protection, asset-combination reduction, and AI-era ad group themes.
6. Greenlane Marketing — Why Your Single Keyword Ad Groups Are Killing Your Google Ads Performance (2025-06-17). https://www.greenlanemarketing.com/resources/articles/why-your-skags-are-killing-your-google-ads-performance (accessed 2026-05-27). Quantifies the SKAG signal-starvation effect on Smart Bidding.