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
The analyst knows the model inside out. The brand team, the performance team and the board don't, and they don't want a dashboard tour. They want to know what happened, why, and what to do about it, in a page and a handful of slides. Writing that summary is often the slowest part of the reporting cycle, so it's a natural job to hand to an assistant that can read the model directly.
Write a 300-word summary of the model, its key findings and what we should focus on next, for our brand and performance teams. Compare 2026-01-01 to 2026-10-31 with the same months of 2025, keep what the model shows separate from your interpretation, and outline a 6-slide deck to present it.
What's behind Meta's jump, and does it hold up against our Meta experiments?
It starts by matching the periods. The current year isn't over, so the assistant compares January to October in both years rather than a partial year against a full one, and says so in the summary. Then it pulls two things: total sales split into base, promotions, weather and media, and spend, incremental sales and ROI by ad platform for both periods.
On our demo account, the headline looks like this:
| January to October | 2025 | 2026 | Change |
|---|---|---|---|
| Total sales | $121 650 061 | $130 873 673 | +7.6% |
| Base sales | $57 315 669 | $58 425 809 | +1.9% |
| Promotion-driven sales | $22 632 862 | $22 174 348 | -2.0% |
| Media-driven sales | $40 930 962 | $50 055 082 | +22.3% |
| Media spend | $13 472 904 | $13 121 336 | -2.6% |
| ROI (MMM) | 3.15 | 3.79 | +0.64 |
Illustrative demo data.
The "what the model shows" part of the summary writes itself from there: sales grew by about $9.2 million, almost all of it media-driven, on slightly less spend. Base demand was roughly flat and promotions did a little less.
The platform breakdown names the drivers. Meta's incremental sales roughly tripled, from $4 230 436 to $13 110 766, on 28% more spend, taking its ROI from 2.52 to 6.12. Google Ads, the biggest platform, improved its ROI from 3.77 to 4.19 on slightly lower spend. On the other side, cuts to TV (-24% spend), direct mail (-14%), Criteo (-50%) and print (-65%) cost incremental sales in each of them.
The follow-up is where a good summary earns trust. Inside Meta, two channels did the heavy lifting:
An ROI jump that size is worth checking before it goes on a slide. The account has conversion lift tests on Meta Sales from 2026, with incremental ROAS between 3.3 and 4.6 (illustrative demo data). That supports the direction, but it's well below the model's 8.68. So the assistant puts it in the summary as a finding to verify, not a headline to celebrate.
The interpretation section then turns this into priorities, clearly labeled as the assistant's reading rather than model output: reconcile Meta Sales with the lift tests, decide what role TV should play at an ROI of 1.42, and look at email, where spend rose 18% at an ROI of 1.37. The 6-slide outline follows the same order (headline, sales decomposition, platform results, the Meta story, priorities, next steps) and flags the last slide as needing data it hasn't pulled yet, such as an optimizer scenario for next quarter.
A strong year can have many causes: more demand, better promotions or better media. Only a model that separates them tells you which one to repeat, and a summary that keeps "the model shows" apart from "we think" lets the audience see which is which.
It also makes the summary safer to share. Drivers that aren't in the data, like pricing or store openings, stay out of the numbers. If you know about them, paste them in as your own context.
Want a summary like this from your own model? Book a demo, or read how to connect Sellforte MCP to your LLM client. The Sellforte MCP launch post covers the basics.
New use cases are added to this library regularly.