Welcome to the Sellforte MCP Use Case Library. Each post takes one question that marketing teams really ask their AI assistant through Sellforte MCP, which is now in general availability and accessible to all marketers worldwide. It strips out anything customer-specific and walks through what the assistant does with it, step by step. We run every question on our demo account so the numbers you see are illustrative, not anyone's real results. The idea is simple: steal the prompts, try them on your own model, and see what you get back. We hope you find this series inspiring!
-Juha, CEO at Sellforte
Retargeting and display platforms are hard to judge. Their own reports usually look great, because they show ads to people who were already close to buying. Then the MMM credits them with much less, and someone in the team asks whether to keep, fix or cut the platform. If you run performance marketing, you've probably had this conversation more than once.
Why does Criteo underperform on ROI, how incremental is it really, and what share of revenue does it drive? Use 2026-05-01 to 2026-10-31.
How does it compare with our other retargeting channels, and is there an experiment we can check it against?
Before it calculates anything, the assistant checks every channel and campaign name that matches the platform. That sounds fussy, but it matters: if a platform shows up under more than one name, a zero under one of them can just mean "not modeled under this name", not "no incrementality". On our demo account, Criteo has three channels (Performance, Display and Video) across two countries. Only the two Performance campaigns had spend in the window, so the rest of the analysis is about those.
Then it lines up three views of the same spend, plus the cost metrics, campaign by campaign:
| Campaign | Spend | ROI (MMM) | ROI (GA4) | ROAS (ad platform) | CPC | CPM | CPA |
|---|---|---|---|---|---|---|---|
| Criteo Performance, DE | $42 101 | 2.88 | 2.90 | 13.04 | $0.36 | $1.96 | $5.46 |
| Criteo Performance, US | $18 525 | 2.77 | 2.90 | 13.04 | $0.36 | $1.96 | $3.83 |
| Criteo total | $60 626 | 2.85 | 2.90 | 13.04 | $0.36 | $1.96 | $4.83 |
| All channels | $7 446 000 | 3.94 | 2.18 | 16.19 |
Illustrative demo data, 2026-05-01 to 2026-10-31.
Three things stand out. The platform reports a return about 4.6 times what the MMM credits. GA4 and the MMM, on the other hand, agree closely (2.90 against 2.85). And the MMM puts Criteo's ROI well below the all-channel average of 3.94. So yes, it's weaker than the rest of the mix, but it isn't a zero. Each dollar still brings back about $2.85 in incremental sales on the demo data.
On share, Criteo took 0.8% of media spend and drove 0.6% of all media-driven incremental sales. Measured against total sales, base and promotions included, its contribution in the window was $172 612 out of $76 900 211, or about 0.2%.
The follow-up puts it next to the other retargeting channels over the same months:
| Retargeting channel | Spend | ROI (MMM) | ROI (GA4) |
|---|---|---|---|
| Criteo Performance | $60 626 | 2.85 | 2.90 |
| Meta Retargeting | $51 672 | 2.08 | 1.76 |
| TikTok Retargeting | $40 614 | 1.59 | 0.59 |
| Pinterest Retargeting | $657 | 2.28 | 0 |
Illustrative demo data.
Inside the retargeting group, Criteo actually comes out on top. As for experiments, the assistant went through the account's 24 incrementality tests and found none on Criteo. The closest evidence is the Meta retargeting holdouts, which measured an incremental ROAS of 1.45 in Germany and 1.7 in the US in November 2025. That's a different platform, so the assistant uses it as context, not as proof. Its suggestion is a holdout test on Criteo itself, for example a geo split, to settle the question.
A retargeting platform's own ROAS answers "did people who saw the ad buy?". The decision you need to make depends on a different question: "would they have bought anyway?". Treat the platform number as the hypothesis, and use the model, and ideally a test, to check it.
On the demo data, the answer isn't "cut it". It's "this is a modest, below-average channel that still pays back, and it's the best of our retargeting options". That's a very different conversation from the one the platform report starts. The data shows where the platform falls short. Why it falls short, whether it's audience overlap with other channels or the people it targets, is still a hypothesis until you test it.
Want to run this on your own retargeting spend? Book a demo, or see how to connect Sellforte MCP to your LLM client. There's more background in the Sellforte MCP launch post.
New use cases keep landing in this library, so come back for more.