Meta has published a number that will be difficult for performance teams to ignore.

Its latest creator-marketing research says campaigns where at least 30% of the creative mix consisted of partnership ads produced a 45% lower median cost per incremental purchase across 20 EMEA conversion-lift studies.

Meta also reports a 35% higher median incremental return on ad spend.

Those are strong numbers. They are also Meta’s numbers.

The analysis covers 20 conversion-lift studies conducted in EMEA between 1 January and 30 June 2025, and Meta reports 92% confidence for the result.

That makes the finding useful as a test hypothesis. It is not an industry-wide benchmark.

The better question for advertisers is whether the effect survives inside their own account.

What Meta actually measured

Partnership ads let brands run advertising using content associated with creators or other partners rather than relying only on conventional brand-owned creative.

Meta’s analysis reports stronger incremental outcomes among campaigns where at least 30% of the creative mix consisted of partnership ads.

Meta-reported measure Result
EMEA conversion-lift studies 20
Partnership-ad creative share At least 30%
Median cost per incremental purchase 45% lower
Median incremental ROAS 35% higher
Reported confidence 92%

The word incremental is important.

This is not simply a comparison of conversions attributed inside Ads Manager. Conversion-lift studies try to estimate additional outcomes caused by advertising by comparing exposed and control groups.

That makes the metric more useful when the question is whether creator advertising generates purchases that would not otherwise have happened.

It still does not make the result universal.

Do not turn 30% into a magic threshold

The tempting interpretation is simple:

Put partnership ads into 30% of the creative mix and acquisition costs fall 45%.

The research does not establish that.

The 30% figure describes the group Meta analysed. It does not prove that 29% fails, 30% suddenly works or 50% performs better.

Creative quality also varies enormously. A weak creator asset does not become effective because it carries the correct format label.

Audience, offer, creator fit, campaign objective, product, placement and brand familiarity can all affect the result.

Treat 30% as a test condition, not a guaranteed optimum.

Test inside normal campaign operations

One useful detail in Meta’s material is that partnership ads can run inside business-as-usual ad sets.

That can produce a cleaner test.

If all creator creative moves into a separate campaign with different budgets, audiences and bidding settings, it becomes harder to tell whether any performance change came from the creative or the surrounding campaign structure.

A practical test can start with:

Baseline

  • standard brand creative;
  • existing audiences;
  • existing optimisation event;
  • normal spend distribution.

Test

  • retain comparable campaign structure;
  • introduce approved partnership ads;
  • target roughly 30% of the eligible creative mix;
  • avoid simultaneous offer or bidding changes;
  • measure incremental outcomes where lift testing is available.

The objective is not to create a perfect laboratory.

It is to avoid changing five things at once and then crediting the creator.

Measure more than CPA

A creator test can look successful in Ads Manager while producing little incremental business.

NEMO would measure four layers.

Layer Metric Purpose
Delivery Spend and impressions by creative type Confirm meaningful creator-ad exposure
Platform efficiency CPA, CTR, conversion rate Measure normal campaign performance
Incrementality Cost per incremental purchase, lift Measure additional outcomes
Business value Revenue, margin, repeat purchase Check commercial value

The difference between attributed CPA and incremental CPA is particularly useful.

Suppose partnership ads show a 20% lower attributed CPA.

That looks positive.

But if conversion lift shows almost no additional purchases, the creative may be collecting conversions that would have happened anyway.

The reverse can also happen. A creator asset may look ordinary under standard attribution while creating more incremental demand.

Use the Partnership Incrementality Gap

NEMO calls one useful diagnostic the Partnership Incrementality Gap.

Compare the percentage change in attributed CPA with the percentage change in cost per incremental purchase.

For example:

  • attributed CPA improves 25%;
  • cost per incremental purchase improves 40%.

The 15-point difference suggests ordinary campaign reporting may be understating the incremental advantage.

Now reverse it:

  • attributed CPA improves 30%;
  • cost per incremental purchase improves only 5%.

The creator ads still look efficient in-platform, but much less of that gain appears incremental.

The gap does not explain why the difference exists.

It tells the team where further investigation is worthwhile.

Run the test market by market

EMEA is a broad label.

Performance in France does not automatically predict performance in Germany, Italy, Poland or the Netherlands.

Creator economics differ. So do language, product mix, brand familiarity, offers and audience behaviour.

European advertisers should initially analyse the test market by market.

Market Partnership creative share Attributed CPA change Incremental CPA change Incremental ROAS change
France Test Measure Measure Measure
Germany Test Measure Measure Measure
Italy Test Measure Measure Measure
Netherlands Test Measure Measure Measure

A regional average can hide a weak local result or an unusually strong creator-market fit.

Measure creator quality, not just creator quantity

Meta’s 30% figure creates another risk.

Teams may start optimising toward the amount of partnership creative instead of the quality of it.

Record:

  • creator;
  • format;
  • opening hook;
  • product;
  • offer;
  • video length;
  • creator-led or brand-led framing;
  • audience;
  • placement;
  • spend;
  • incremental outcome.

After enough tests, the useful question becomes more specific than “Do partnership ads work?”

It becomes:

Which creator and creative combinations generate incremental purchases for this product in this market?

That is a much better basis for scaling.

The 92% confidence figure matters

Meta reports 92% confidence for the EMEA analysis.

Many controlled experiments use 95% as a conventional statistical threshold.

That does not make 92% meaningless. It does mean advertisers should resist turning the headline result into a guaranteed performance claim.

A better interpretation is:

Meta has produced evidence strong enough to justify a test, not strong enough to replace one.

The distinction matters when a platform publishes research about advertising products sold by the same platform.

Vendor research can be useful. It should still be reproduced inside the advertiser’s own environment before becoming budget policy.

What European teams should do now

Start with one market or product line that has enough conversion volume to measure properly.

Introduce partnership ads alongside normal business-as-usual creative.

Track delivery so the creator assets receive enough exposure to make the comparison meaningful.

Then compare ordinary attributed efficiency with incrementality. If the account can run a conversion-lift study, use it.

Repeat the test before scaling.

One successful creator is evidence about one creator. Several repeatable tests across products or markets begin to tell you something about the channel.

Meta’s 45% result is useful because it gives advertisers a concrete hypothesis.

The real advantage comes when a brand can replace that number with one of its own.

Sources