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
If you run an omnichannel retail business, someone will eventually ask: "Does our advertising actually drive the stores, or just the web shop? And is it bringing in new customers, or selling to people who'd have bought anyway?" It's usually the marketing or e-commerce manager who has to answer, and a single total ROI doesn't help much.
For the last 12 months (2025-11-02 to 2026-11-01), all markets: how much of our sales comes from media, split by online vs stores, new vs returning customers and product category? State the scope you used.
Which media groups lean most toward new customers, and what's their ROI on new customers alone?
The assistant first splits total sales into base, promotions, weather and media. Over the 12 months, total sales come to 172 567 356, and media drives 66 901 275 of it, 38.8%. The base is 43.6% and promotions 17.5%.
Then it splits media's part by sales channel: 54 519 894 lands in e-commerce and 12 381 381 in stores. One thing it points out straight away: the base and promotions aren't split by sales channel in the model, only media is. So you get "where does media's contribution land", not "what share of store sales is media".
Next, it crosses customer type with sales channel:
| Media-driven sales (illustrative demo data, USD) | E-commerce | Stores | Total |
|---|---|---|---|
| New customers | 24 697 816 | 3 242 075 | 27 939 891 |
| Returning customers | 29 781 191 | 8 771 057 | 38 552 248 |
| Total | 54 479 007 | 12 013 132 | 66 492 139 |
About 42% of what media drives comes from new customers. Online, it's closer to 45%. In stores it's about 27%, so in this demo business, media's store effect is mostly about bringing existing customers back.
You'll notice the media total here (66 492 139) is a little different from the 66 901 275 above. The assistant explains why: the first split uses same-period uplift, which adds up to total sales, and this one uses backtraced uplift, the basis for ROI. Both are right for what they measure.
The product category split came back as a single "Overall" row on the demo account, because its sales data doesn't carry categories. The assistant says so rather than inventing a split. On an account whose sales data has product categories, the same question returns one row per category.
For the follow-up, the assistant splits each media group's incremental sales into new and returning customers, then filters the ROI to new customers only.
| Media group (illustrative demo data, USD) | Spend | New-customer sales | Share new | New-customer ROI |
|---|---|---|---|---|
| Paid Social | 3 216 237 | 9 482 796 | 54% | 2.95 |
| Paid Video | 663 580 | 1 522 458 | 51% | 2.29 |
| Cross-network | 5 397 867 | 10 933 221 | 46% | 2.03 |
| TV | 1 690 984 | 1 307 295 | 56% | 0.773 |
| 1 737 692 | 664 656 | 19% | 0.382 | |
| Direct Mail | 2 005 714 | 206 571 | 2% | 0.103 |
The assistant is careful about what that last column means. Spend has no customer-type split, so new-customer ROI is new-customer sales divided by the channel's full spend. It's a partial ROI. For each channel, the new and returning ROIs add up to the overall one.
Read that way, the table tells a clear story. Paid Social both leans toward new customers and earns the most per dollar from them. TV leans new too, but at a low return. Direct Mail and Email are almost entirely about returning customers, which is fine as long as you're not counting on them for growth.
Platform reports tend to make the channels that catch people already on their way to buy look best. The real question is whether media brought in anyone new, and that needs a model that separates the baseline from what media added. Splitting media's contribution by sales channel and customer type turns "media works" into something you can act on: which channels feed the stores, and which ones grow the customer base.
Two limits to keep in mind. The splits depend on what your sales data carries, and cost per new customer can't be read directly, because spend isn't split by customer type.
Curious what your own split looks like? Book a demo, or check how to connect Sellforte MCP to your LLM client. More use cases are added to this library all the time.