Ask Sellforte MCP: How do our own campaign groups perform?
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
Brand, prospecting, retargeting: grouping campaigns your own way
Most teams have their own way of slicing media: brand versus tactical, prospecting versus retargeting, or campaign families coded into the campaign name. The model and the dashboard don't know about those groups, so a marketing or media manager ends up exporting campaign rows and rebuilding the split in a spreadsheet every time someone asks.
What you ask
Treat campaigns whose names contain "Brand", "Prospecting" or "Retargeting" as three groups, and everything else as Other. Compare spend, incremental sales and ROI per group by year for 2025 and 2026 so far, and list the campaigns in each group.
Which campaigns ended up in Other that look like they belong to one of the groups?
What Sellforte MCP does
We ran this on our demo account, where campaign names follow a platform_campaign_market pattern, such as Meta Ads_Meta Prospecting_DE. All figures are illustrative demo data in USD.
Reads the campaign list first
Before grouping anything, the assistant pulled every campaign with spend between 2025-01-01 and 2026-10-31. That's 77 campaigns. Seeing the real names up front matters, because your rules are only as good as your naming.
Turns each rule into a name filter
Instead of downloading every row and adding them up itself, the assistant passed each rule to Sellforte as a "name contains" filter, so the query layer did the sums and worked out ROI. It also reported exactly which campaigns each rule caught:
- Brand (5 campaigns): Google and Bing brand search in both markets, plus a Performance Max brand campaign in the US.
- Prospecting (6 campaigns): Meta, TikTok and Pinterest prospecting in both markets.
- Retargeting (6 campaigns): Meta, TikTok and Pinterest retargeting in both markets.
- Other: everything else, queried as "all campaigns except those 17".
Returns the table by year
2026 covers 2026-01-01 to 2026-11-01 only, so its spend and sales aren't comparable to a full year. ROI is.
| Group | Spend 2025 | Incr. sales 2025 | ROI 2025 | Spend 2026 (partial) | Incr. sales 2026 (partial) | ROI 2026 |
|---|---|---|---|---|---|---|
| Brand | 1 024 012 | 2 031 874 | 1.98 | 536 314 | 1 247 870 | 2.33 |
| Prospecting | 1 269 001 | 4 676 809 | 3.69 | 583 612 | 3 069 162 | 5.26 |
| Retargeting | 559 154 | 1 102 559 | 1.97 | 235 051 | 444 079 | 1.89 |
| Other | 15 008 596 | 51 415 893 | 3.43 | 11 808 270 | 45 168 277 | 3.83 |
| Total | 17 860 763 | 59 227 135 | 3.32 | 13 163 246 | 49 929 387 | 3.79 |
The four groups add up to the total, which is a quick sanity check that nothing fell through the cracks or got counted twice.
The story is pretty clear. Prospecting is the strongest of the three named groups and got better: ROI went from 3.69 in 2025 to 5.26 so far in 2026. Brand improved a little, from 1.98 to 2.33, but still sits well below the total. Retargeting is flat at around 1.9 and is the weakest group in both years.
Flags what the rules missed
The follow-up is where this gets useful. Looking back at the campaign list, the assistant spotted Meta Remarketing and Pinterest Remarketing campaigns in Other. They do the same job as retargeting, but the word in the name is different, so the rule didn't catch them. On demo data they're small, but in a real account a naming slip like that can quietly skew a group.
The fix is just a tighter rule: add "Remarketing" to the retargeting group and rerun.
Why it matters
Your groupings are how your team actually makes decisions. If the budget debate is "brand versus everything else" or "how much do we put into prospecting", you need the numbers in those terms, not in whatever dimensions the model happens to have.
It's the same idea as rebuilding a funnel role from campaign names before comparing years: the classification has to be right before any trend means anything. Here, a few lines of rules turn 77 campaigns into four groups, and the matched-campaign list lets you check the rules did what you meant.
One thing to keep in mind: the groups only exist for that conversation. If you want the same split every month, keep the rules in a saved prompt and reuse it.
Try these follow-ups
- Add "Remarketing" to the retargeting group and redo the table.
- Show the same groups by quarter, with ROI for each group as a trend next to the total.
- Split the Brand group by country. Is the US brand ROI different from Germany's?
- Within Other, which ten campaigns have the most spend?
- Treat everything except brand campaigns as tactical, and compare the two by month this year.
Related
- Marketing mix modeling
- Sellforte Performance
- Sellforte dashboard
- Ask Sellforte MCP: Can I just get the campaign table?
Try it yourself
Curious how your own campaign families stack up? Book a demo, or if you already use Sellforte, here's how to connect Sellforte MCP to your LLM client. We add new use cases to this library regularly.
