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
You switched on a new campaign type in one market a few weeks ago, and now someone's asking whether it's working. Total sales are up, which feels like a yes. But a lot of things changed in those weeks, and "sales went up after we launched" isn't the same as "the launch lifted sales". Growth and market managers running pilots need the second answer, ideally before the next budget call.
We started Google Shopping in Germany on 2026-09-13. Do we see an uplift in sales compared with the seven weeks before? Show spend, incremental sales and ROI, and tell me the last date the model data covers.
How much of Germany's sales change in those weeks came from Shopping, and how much from base demand, promotions and other media? How does Shopping compare with our other Google channels there?
Before any numbers, the assistant checks how far the model goes. On the demo account, MMM results run to 2026-11-01, refreshed on 2026-11-02. So it compares seven full weeks after the launch (2026-09-13 to 2026-10-31) with the seven weeks before (2026-07-26 to 2026-09-12), and says so.
The Shopping campaign appears in the data from the week of 2026-09-13, with no spend in the seven weeks before. Week by week it's steady (illustrative demo data, USD):
| Week starting | Spend | Incremental sales | ROI (MMM) |
|---|---|---|---|
| 2026-09-13 | 5 996 | 11 616 | 1.94 |
| 2026-09-20 | 7 068 | 15 717 | 2.22 |
| 2026-09-27 | 7 104 | 15 920 | 2.24 |
| 2026-10-04 | 6 684 | 15 711 | 2.35 |
| 2026-10-11 | 6 697 | 13 428 | 2.01 |
| 2026-10-18 | 6 693 | 12 528 | 1.87 |
| 2026-10-25 | 6 729 | 16 332 | 2.43 |
In total: 46 971 spent, 101 252 incremental sales, an ROI of 2.16. It pays back, and it hasn't faded since week one.
Then the bigger picture. Germany's total sales rose 28.3% between the two windows, about 3 312 000 more. The model splits that into its parts:
Shopping's 101 252 sits inside that media line. So yes, sales went up after the launch, but most of the rise came from promotions and underlying demand, not the new campaign. The assistant adds that the media figure here is same-period uplift, while the campaign numbers use the ROI basis, so the two don't add up exactly by design.
Last, it lines Shopping up against the other Google Ads channels in Germany over the same seven weeks (illustrative demo data, USD):
| Google Ads channel (DE) | Spend after | ROI before | ROI after |
|---|---|---|---|
| Performance Max | 396 872 | 5.12 | 4.79 |
| Search Generic | 61 014 | 5.71 | 3.96 |
| YouTube | 54 813 | 4.36 | 4.71 |
| Shopping | 46 971 | – | 2.16 |
| Search Brand | 23 372 | 4.13 | 3.42 |
Shopping has the lowest return of the group. Spend went up on every other channel too, by 12% to 186%, so the launch didn't happen in isolation. Several channels' ROI slipped as their spend grew, which a before-and-after view can't untangle from the launch.
The tempting read is "sales are up 28%, the pilot worked". The model says something more useful: Shopping brings in about 2.16 for every dollar, steadily, which is positive but the weakest return among your Google channels in that market. That points to keeping a close eye on it, rather than scaling it on the strength of the total sales number.
And a before-and-after comparison on the MMM isn't a controlled experiment. Promotions nearly tripled in the same weeks, and other Google spend rose too. If the decision is big, the honest next step is a lift test that holds part of the market out, so the result doesn't depend on everything else standing still.
Running a pilot of your own right now? Book a demo, or follow the guide on connecting Sellforte MCP to your LLM client. There's more background in the launch post.
Keep an eye on this library: new use cases arrive regularly.