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
Mapping two years of channels, tests included
Ask a team which media channels they've run over the last two years and you'll usually get the big ones right away. The short pilots, the one-quarter experiments and the format someone tried before the holidays are much harder to remember. If you're new to the team, planning next year or trying to understand what goes into your marketing mix model, you need the full list, with dates.
What you ask
List all media channels used in 2025 and 2026. For each, give the first and last month with spend, the number of active months and total spend. Flag channels active for fewer than six months as tests or one-off campaigns.
For those short tests, what did they return? Show spend, incremental sales and ROI next to the ongoing channels.
What Sellforte MCP does
The assistant starts by checking which datasets the connection exposes and how far back the data goes. On our demo account the model data runs from the end of December 2024 to 2026-11-01, so it uses January 2025 to October 2026 as the window.
Then it pulls monthly spend by media channel from the media performance dataset. That's one query, about 400 rows: every channel, every month with spend. From there the assistant works out, for each channel, the first and last month with spend and how many months were active. It doesn't guess the threshold for a test. It uses the six months you gave it and says so.
The follow-up is a second query on the same dataset: total spend, incremental sales and ROI per channel over the whole window, straight from the model.
The short runs
On the demo account there are 26 media channels with spend in the window. Six of them were active for fewer than six months:
| Media channel | First month | Last month | Active months | Spend (USD) | ROI (MMM) |
|---|
| Paid Video Retargeting | 2025-09 | 2025-12 | 4 | 115 775 | 2.27 |
| Performance Max Brand | 2026-09 | 2026-10 | 2 | 28 955 | 3.32 |
| Paid Social Remarketing | 2025-01 | 2025-08 | 5 | 18 669 | 2.31 |
| Paid Social Leads | 2025-11 | 2026-07 | 4 | 17 289 | 3.28 |
| Paid Social Engagement | 2025-11 | 2026-03 | 5 | 6 290 | 3.87 |
| Smart | 2025-10 | 2025-12 | 3 | 544 | 2.89 |
Illustrative demo data. Demo account figures are in USD.
For scale, the whole media mix spent 30 982 099 USD over the same 22 months at an ROI of 3.52. The biggest channels were Performance Max (6 677 509 USD, ROI 4.56), Direct Mail (3 980 000 USD, ROI 4.29) and TV (3 343 421 USD, ROI 1.54).
The patterns a simple threshold misses
Here's where reading the monthly rows pays off. A few channels clear the six-month bar but still tell a story you'd want to know:
- Paid Social Advantage+ first shows up in 2025-12, then runs every month from 2026-03 onwards. That's a new channel being scaled, not a test.
- Paid Shopping spent 634 567 USD in 2025-01, wound down through 2025, had no spend from 2026-01 to 2026-08 and came back in 2026-09.
- Display Retargeting has no spend after 2026-03.
- TV runs in flights: 13 active months out of 22, with gaps in between.
And one of the "tests" probably isn't. Performance Max Brand started in 2026-09, two months before the data ends. A short history at the end of the window means "new", not "tried and dropped". The assistant can only see spend, so it's worth asking it to keep those two cases apart.
Why it matters
A channel inventory sounds like housekeeping, but it's the first thing you need before you plan, brief an agency or question a model result. Short tests are easy to forget, and they're also the channels with the least data behind them. A two-month run gives the model much less to learn from than two years of steady spend, so it's worth knowing which results rest on a handful of months before you lean on them.
It also helps you spot what's missing. The list only covers channels that feed into the model. If something you ran doesn't appear, that's a data question to raise, not a result to report.
Keep in mind that "test" here is inferred from how long a channel had spend. The model doesn't know which channels you planned as tests. You do, so treat the flag as a starting point for the conversation.
Try these follow-ups
- Which ad platform and campaigns made up each short test?
- Mark channels that started in the last three months as new rather than tests.
- Which channels stopped spending in the last six months, and what was their ROI before they stopped?
- Show the same list by country, so I can see which markets ran which tests.
- Return the list as a table I can sort by spend.
Related
Try it yourself
Want to see this on your own data? Book a demo. Already a customer? Here's how to connect Sellforte MCP to your AI assistant.
New use cases land in this library regularly, so check back soon.
