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
Before you build a report, feed a warehouse or plan a new workflow on top of Sellforte, you need to know what's actually there. Which metrics, which breakdowns, how far back, and at what grain? Analysts, BI leads and group marketing teams ask this early, usually when they're deciding whether the connection can replace a manual export. Finding out the hard way, halfway through a build, is expensive.
List the datasets, metrics and dimensions Sellforte exposes for marketing mix modeling on my account, the lowest time grain and the earliest date available, and whether campaign name and campaign objective are included. Test one small pull to confirm.
Which other domains can I query on this account?
The assistant starts with a schema discovery call for the marketing mix modeling domain. On our demo account, that came back with three datasets:
| Dataset | What it holds | Example metrics |
|---|---|---|
| Media performance | Spend and results by channel, campaign and country | Spend, impressions, clicks, CPM, CPC, incremental sales, ROI, plus GA4 and ad platform figures for comparison |
| Uplift | Media-driven sales split by sales-side breakdowns | Incremental sales by customer type, product category or sales channel |
| Sales decomposition | Total sales split into components | Base, promotions, weather and media |
It also listed the dimensions for each dataset: ad platform, advertising channel, campaign name, campaign objective, country, media type, media group and media channel, plus customer type, product category and sales channel on the sales side.
A detail the assistant picked up from the schema and passed on: in the media performance dataset, the sales-side breakdowns work as filters only. You can narrow incremental sales to, say, one sales channel, but spend stays at the channel's full total, because spend has no sales-side split. That's exactly the kind of thing you want to know before you design a report around it.
The schema answered the time questions directly. Time breakdowns go from total down to year, quarter, month, week and day. On the demo account, data runs from 2024-12-30 to 2026-11-01, last updated on 2026-11-02, and figures are in USD.
Schemas can promise more than the data delivers, so the assistant ran two small pulls. The first asked for spend and incremental sales by campaign objective, by day, for 2026-10-29 to 2026-10-31 (illustrative demo data, USD):
| Date | Campaign objective | Spend | Incremental sales |
|---|---|---|---|
| 2026-10-29 | Revenue | 31 705 | 123 138 |
| 2026-10-29 | Clicks | 2 123 | 6 950 |
| 2026-10-29 | Conversions | 312 | 1 360 |
| 2026-10-31 | Revenue | 34 151 | 110 807 |
| 2026-10-31 | Clicks | 1 508 | 4 826 |
| 2026-10-31 | Conversions | 259 | 969 |
So daily data by campaign objective is there. The second pull asked for campaigns in the very first month, 2024-12-30 to 2025-01-31, with a deliberately tight row limit. Sellforte MCP declined to return a partial list and reported that the full answer has 46 campaign rows, which confirms campaign-level detail goes back to the start of the data. It's a nice touch: rather than quietly cutting the list short, it tells you how big the real answer is.
The follow-up showed what else the account is served. On the demo account that's promotions, optimizer scenarios, campaign-level budget and bid recommendations, campaign performance history, bid change history and experiments. Not every account has every domain, so this is worth asking on yours.
A data inventory sounds dull, and it's the step people skip. It's also what decides whether a report or integration works. The answer you get here is a clear go or no-go, with the grain known before anyone writes a line of code or builds a dashboard.
The test pull is the part to keep. A schema tells you a field exists. A three-day pull tells you it's populated, at the grain you need, with numbers that look sensible. And for some fields the honest answer is "not available". That's a useful answer too, as long as you get it before the build, not after.
Book a demo to see what Sellforte MCP exposes for a model like yours, or read how to connect Sellforte MCP to your LLM client. New use cases land here regularly.