12 paid-ad results, each credited and linked to the source.
All 168 campaign structure results in this hub report an improvement, and the 48 that give a percentage drop in a cost metric (CPA, CAC, CPL or similar) show a median cut of 38%.
Treat that number with care: 100 of the 168 are vendor-reported case studies, and only 16 are split tests.
The changes that repeat most are moving to an automated campaign type (Smart+, Performance Max, Advantage+ or GMV Max) and merging many small campaigns into a few larger ones.
A result lands here when the main change was how campaigns, ad sets and budgets were organised, not the ads themselves.
That includes merging campaigns, splitting one campaign into themed ad sets, switching between CBO and ABO, moving to an automated campaign type, and segmenting by brand vs non-brand, location or funnel stage.
Many results also changed other things at the same time.
51 of the 168 are also tagged with audience changes, 27 with budget changes and 25 with creative volume.
You can browse all results to compare structure with other levers.
Every one of the 168 results is marked as an improvement.
That says more about which results get published than about how reliable restructuring is.
Agencies and ad platforms share wins, and none of the results in this set report a loss.
Of the 48 results that report a percentage drop in a cost metric (we took the first cost figure in each headline and left out one unsourced claim), the cuts run from 3% to 96%, with a median of 38%.
The 9 split tests among them have a lower median of 32%. See how this change ranks against others.
| Evidence type | Results | Vendor-reported |
|---|---|---|
| Before/after | 99 | 47 |
| Campaign result (no comparison) | 39 | 30 |
| Split test | 16 | 12 |
| Lift study | 9 | 9 |
| Claim or snapshot | 5 | 2 |
The largest result graded as a split test is vendor-reported.
TikTok says Marks & Spencer cut purchase CPA 94% moving from manual campaigns to Smart+ with creator content and catalog ads, with conversion rate up 44%.
That change mixed new creative with the new campaign type, so structure alone did not drive it.
Among agency results, Flighted merged three competing Meta campaigns into one Advantage+ Shopping campaign with cost caps for K Republiq: ROAS 1.62 to 3.1, CPA $40 to $27. A new landing page launched at the same time.
The cleanest example of splitting rather than merging comes from @AshvinMelwani: Obvi broke a 50-ad CBO into 5 ad sets and reported NC-ROAS +55.65% (1.79).
It is a before/after with no spend shared.
TikTok has the most results (65), then Meta and Google with 58 each and YouTube with 11. 21 results span more than one platform.
The source mix differs a lot by platform:
86 results name an ad format. Video ads (42), UGC ads (35), partnership ads (33) and catalog ads (25) lead.
Catalog ads show up mostly in Smart+ and Advantage+ shopping setups, where the platform picks products from a feed.
Ecommerce (34), apparel (30), beauty (16) and accessories (13) make up most of the hub.
Finance, apps, SaaS and food and beverage have 8 each, so results outside retail are thin.
Fewer, bigger campaigns. About half the results name an automated campaign type in the headline or title, and many more describe merging campaigns.
Google says Catawiki cut 100+ Search campaigns to about ten and got almost 40% more conversions at the same cost of acquisition (vendor-reported).
For Meta, compare advantage plus vs manual campaigns.
Structure follows budget control. CBO and ABO both appear among the winners. @SamPiliero shared community members' posts about an ABO account at ROAS 5.53 on $50k and a CBO prospecting campaign with 46 sales in 7 days, so neither setup wins on its own.
Both are self-reported, and results like these vary and are not typical. See cbo vs abo.
Segmenting where intent differs. On Google, many agency wins split brand from non-brand, or segment by location or product margin.
KlientBoost's Everbloom Baby rebuild on Facebook cut CPA 60% while spend tripled, but CRO tests ran alongside it.
No result in this hub is marked as lost or flat, so there is no failure data to learn from directly.
The caveats come from how the wins were measured:
The better-measured results are rarer. Power Digital's YoungLA work used geo holdout tests to guide budget moves and reports incremental ROAS +220% and CAC −32% after cutting 50+ Meta campaigns to three priorities, though those headline figures are a before/after of the whole program, which also added CTV and new markets.
Restructuring probably helps least on accounts that are already simple or too small to give one campaign enough conversions.
The same discipline applies to how to run a creative test.
If you need new ads to fill a consolidated campaign, swipe files for ideas can speed that up.
No single structure wins in these results. Meta winners include one consolidated CBO, ABO accounts and Advantage+ Shopping setups.
The repeat pattern is fewer campaigns with enough budget and conversions each, for example K Republiq merging three campaigns into one ASC (ROAS 1.62 to 3.1).
It often did in these results, but most are before/after comparisons where creative or landing pages changed too.
Across the 48 results in this hub that report a percentage cost drop, the median cut was 38%.
No losing results were published, so expect some tests to fail.
Results here lean towards fewer campaigns with Smart Bidding and Performance Max, plus separate brand and non-brand campaigns.
Google says Catawiki cut 100+ Search campaigns to about ten and got almost 40% more conversions at the same cost of acquisition (vendor-reported).
16 of 168 are split tests and 9 are lift studies.
12 of the split tests and all 9 lift studies are vendor-reported.
The remaining 143 are before/after comparisons, campaign results or claims.