Meta White Paper: 30% Partnership Ads Saw 45% Lower Cost per Incremental Purchase in EMEA Lift Studies

Meta published a 54-page creator marketing white paper on 7 October 2026 claiming that advertisers who put at least 30% partnership ads into their business-as-usual campaigns recorded a 45% lower median cost per incremental purchase across 20 conversion lift studies run by EMEA advertisers. The figure is Meta's own, company-supplied analysis, and every study behind it was run by an advertiser in EMEA. This summary is based on PPC Land's detailed reading of Meta's paper, published on 11 October 2026. For brands in Thailand that pay KOLs and then boost their posts, the most useful part of the paper is not the headline number but the three-stage test plan at the end.

What Meta published on 7 October 2026

According to PPC Land, the material came in two parts. The first is a Meta for Business landing page filed under "Inspiration", dated 7 October 2026 and headed "Together creates better: building for a creator-first future." The second is the downloadable PDF. Its cover carries a different title, "Cracking the creator code - The new science of creator campaign measurement", and PPC Land reports that the file's metadata shows it was created on 7 October and modified on 8 October 2026. Anyone searching for the document later may find it under either name.

The foreword is written by Derya Matras, Meta's Vice President for EMEA. Her argument, as quoted by PPC Land, is that measurement is usually the last thing the industry gets to, and that "the winners will be the ones who get there first." The paper organises its advice into five moves: Confront the shift, Convert with creators, Craft the creative, Connect organic and paid, and Confirm the payback. PPC Land's verdict is that the landing page reads like a manifesto while the PDF is mostly a measurement manual.

The 45% claim and how it was measured

The headline number comes from a meta-analysis of 20 conversion lift studies run between 1 January and 30 June 2025 by advertisers in EMEA. In each study, one cell ran the advertiser's existing media strategy. The other cell was the same strategy with at least 30% of its assets supplied as partnership ads creatives, the format that lets a brand run an ad from a creator's handle. Meta reports that the treatment cells recorded a 45% lower median cost per incremental purchase and a 35% higher median incremental ROAS.

Three details from PPC Land's reading matter for anyone planning to quote the number:

• The stated confidence level is 92%. That is below the 95% threshold most academic and many commercial tests use, and the paper does not comment on it.
• On page 24 the same result is worded differently: incremental conversions were "45% cheaper" and the threshold is "over 30% of campaign assets" rather than "at least 30%". The two passages appear to describe the same dataset.
• The figures are medians across 20 conversion lift studies, so they describe the middle result, not what every advertiser saw.

Meta presents the 30% threshold as a planning rule. PPC Land quotes the paper as saying it is "a number to brief against, rather than a judgement call." Note that the threshold is a share of assets in the test cell, not a share of spend. Another of Meta's figures, covered below, uses a share of spend instead, so the two should not be mixed.

The other numbers in the paper, and their samples

The table below lists the main performance figures in the paper as PPC Land reported them, with the sample and period behind each one. All of them except the IPA index come from Meta's internal analyses.

The 63% new-reach figure has a geographic oddity that PPC Land flagged. The 10 advertisers were "based in EMEA", but the footnote says their ads were delivered predominantly in North America, with the United States accounting for about 43% of spend, Europe around 13% and APAC about 10%. The analysis covered 684 ad sets where more than 15% of reach came from both partnership and non-partnership ads.

Of the figures PPC Land reports, the 19% figure is the only one whose sample includes APAC tests, and that data is more than four years old. None of the reported figures is specific to Thailand or to Southeast Asia.

Why Meta builds the argument on the IPA study

The paper's core case rests on one independent source: the IPA study "Beyond engagement - understanding influencer payback", published in October 2025. According to the white paper, creators scored 99 on short-term ROI against an all-media average of 100, "performing on par with linear TV while accounting for 4.5% of short-term sales." Over the long term the same dataset put creators at 151, with a long-term multiplier of 3.35, ahead of linear TV at 3.27 and paid social at 1.98.

That gap between 99 and 151 is the problem Meta is trying to solve. Most budgets are set on short-term results, and on a short-term view creators look average. Meta's framing, quoted by PPC Land, is that "the gap isn't belief or budget. It's infrastructure: how creators are planned, briefed, bought and measured." The landing page cites 220 campaigns and 144 brands for the IPA dataset. The PDF cites the IPA report by title and date but gives no sample size.

Some statistics in the paper have no source at all. PPC Land points to the line that "70% of content on Meta technologies is from accounts that people don't follow or aren't friends with", which carries no footnote, and to a 66% reach claim on page 24 that is also unsourced. Other market figures are footnoted to outside research firms.

What L'Oreal and Unilever contributed

The most practical section, in PPC Land's view, comes from Vasileios Kourakis, Global Marketing Effectiveness Director at L'Oreal. "First, we fixed the data," he writes. L'Oreal removed duplicate entries, mapped every creator cost down to the individual piece of content, reconciled those costs with finance, and applied consistent naming and tagging. It also separated sponsored content, organic advocacy and boosted advocacy, because each behaves differently and needs to be measured differently.

L'Oreal then used three layers: media metrics for fast reads, incremental lift experiments in the middle, and marketing mix modelling for sales impact. According to Kourakis, MMM work showed that campaigns using brand plus creator content delivered 35% higher ROI. The footnote limits that to L'Oreal Luxe in the UK and US, covering 1,150 campaigns between January 2022 and December 2023.

Unilever's contribution is the Vaseline Verified campaign, which invited more than 450 influencers to co-create content around some 3.5 million Vaseline "hacks" found online. The paper reports 136 million views and a 43% lift in sales, but PPC Land notes that the sales figure is given without attribution.

The setup advice Meta gives

Alongside the evidence, the paper describes Meta Creator Marketing Hub, which brings creator discovery, content sourcing, permissions and partnership ads activation into one place. Creator Marketplace now lists Facebook creators next to Instagram creators, and the paper describes more than 5 million creators as discoverable. Two new Meta Business Suite permissions, "Creator management" and "Creator content", are meant to give creator and media teams access without unrelated rights.

The operational guidance is specific:

• Ask creators for account-level permissions, so the brand can run any of the creator's posts that tag it and create new partnership ads without an existing organic post.
• Do not run partnership ads on their own. Put them in the same ad set as business-as-usual ads, or in the same campaign with Advantage campaign budget switched on, so delivery can shift between the brand and the creator.
• Use Advantage+ placements, Reels and a dynamic header that switches between creator and brand handles.

Meta also lists three common blockers. Agencies need ad account access plus ads access to both the brand's Instagram account and Facebook Page. If a creator's Facebook and Instagram accounts are linked, both must meet eligibility rules. And an advertiser needs a Facebook Page to run partnership ads at all, while creators without one will not deliver to Facebook placements.

The three-stage test ladder

The paper closes with a three-stage testing plan. Each test is meant to run for at least one full conversion cycle.

1. Early explorer. Compare the current strategy without creators against the same strategy plus creators delivered as partnership ads. Run it for 3-4 weeks for a brand lift read or 3-8 weeks for a conversion lift read.
2. Growth optimiser. Compare business-as-usual creator selection against content recommended by Partnership Ads Hub, a test the document says is available on Instagram only. The alternative at this stage is business-as-usual against macro and micro creators.
3. Sophisticate adopter. Compare catalog ads with and without partnership ads.

PPC Land's comment on the plan is that the design is conventional, but it is by construction a test run inside Meta's own lift tooling and measured by the platform that sells the inventory.

What the paper does not tell you

The limits are worth stating plainly before anyone puts the 45% figure in a plan. The specific, actionable numbers (30% of assets, 45% cheaper, 35% higher incremental ROAS, 63% new reach) are Meta's own, drawn from samples of 10 to 20 advertisers. Two of the confidence levels sit below conventional thresholds, and one dataset dates back to 2021. PPC Land also points out that partnership ads are a large business line for Meta: when Meta reported first-quarter 2026 results, partnership ads had reached a $10 billion revenue run-rate.

The 45% claim has no APAC or Thai results behind it, because its 20 studies were run by EMEA advertisers. Meta itself frames the paper as a way to prove the case "not with someone else's benchmark, but with a test run in your own business." That is the right way to read it: as a hypothesis to test, not a result to assume.

What this means for Thai marketers

This section is analysis, not part of Meta's paper. A common setup in Thailand is to pay a KOL for a post on their own account and then boost it, or to run creator content in a separate campaign from the brand's always-on ads. Meta's guidance points the other way: run creator content as partnership ads inside the same ad set or campaign as the brand's existing ads, so delivery can adapt between brand and creator, and then measure the difference with a lift test.

A practical order of work, based on the paper's own advice, would be:

• Clean the cost data first, as L'Oreal did. Map each KOL fee to the specific piece of content it paid for, and keep sponsored posts, organic advocacy and boosted posts as separate lines.
• Fix the access problems before the campaign starts: a Facebook Page for the brand, ads access for whoever runs the account, account-level permissions from each creator, and eligible linked accounts.
• Run stage one of the ladder on your own account. Use at least 30% partnership ads assets in the test cell, so the result is comparable with Meta's threshold, and let it run for a full conversion cycle.
• Judge the result against your own baseline cost per incremental purchase, not against the EMEA median.

Teams that manage both brand and creator activity on Meta can find more on campaign structure on our Facebook ads service page, and on how creator content fits with other paid social channels on our social ads page.

FAQ

Does Meta's 45% result apply to advertisers in Thailand?

Not directly, because the 20 conversion lift studies behind it were run by EMEA advertisers between January and June 2025. The paper contains no Thai or Southeast Asian results for this claim, so the only way to know whether the effect holds in Thailand is to run the same kind of lift test on a Thai account.

What does "at least 30% partnership ads" mean in practice?

It means at least 30% of the assets in the test cell were partnership ads creatives, according to PPC Land's reading of the paper. It is a share of assets, not a share of budget. A separate Meta figure, the 71% intent lift, used a threshold of at least 20% of cell spend, so the two thresholds measure different things.

How long should a creator lift test run?

Meta suggests 3-4 weeks for a brand lift test and 3-8 weeks for a conversion lift test, and says every test should cover at least one conversion cycle. Our reading is that a product with a long purchase cycle will need the longer end of that range.

Is the evidence in the white paper independent?

Mostly not. The long-term ROI index of 151 comes from the IPA, an independent trade body, but the 30%, 45%, 35% and 63% figures are Meta's own analyses of Meta campaigns. PPC Land also notes that the 45% result was reported at 92% confidence, below the usual 95% threshold.

Do I need a Facebook Page to run partnership ads?

Yes, according to the paper, an advertiser needs a Facebook Page to run partnership ads at all. Creators without a Facebook Page will not deliver to Facebook placements, and if a creator's Facebook and Instagram accounts are linked, both must meet the eligibility rules.

The bottom line

Meta's paper gives advertisers a clear rule of thumb and a clear test plan, and both are worth using. The 45% figure is a starting hypothesis from 20 EMEA advertisers, measured by Meta. A Thai brand that already spends on KOLs can test what partnership ads are worth in its own account, running each test for at least one conversion cycle as Meta's plan suggests. If you want help setting up that first lift test, talk to the Relevant Audience team.