Rumorly vs a Make + ChatGPT scenario
A scenario turns free text into a dropdown value. Rumorly turns it into a named source with revenue next to it, and keeps getting better at your answers.
The DIY version is real, and sometimes enough
The common setup: a form or CRM trigger fires, Make (or Zapier) sends the answer to ChatGPT with a prompt like "pick one of these 8 channels", and the result goes into a dropdown property in HubSpot or Attio. There are public walkthroughs of exactly this, and it takes an afternoon.
If all you want is a rough channel mix and you'll look at it once a quarter, keep it. It's cheap and you own it.
Where it stops
It picks a bucket, not a name. "Podcast" is a bucket. "Lenny's Podcast, the PLG pricing episode" is a decision: book that show again. Getting names out of a prompt is possible, but then you need a dictionary so "Lenny", "Lennys pod" and "Lenny Rachitsky" land on one entity, and a place to approve new ones. That's the review queue in Rumorly.
Corrections don't stick. When you fix a wrong value in the CRM, the scenario doesn't learn it. In Rumorly an override is stored as a labelled example and the dictionary grows, so the same answer is right next time.
There's no revenue. A dropdown in the CRM still has to be joined to billing. Rumorly reads Stripe with a read-only key, turns active subscriptions into monthly revenue (trials out, discounts applied, annual divided by twelve, currencies converted), matches customers to respondents, and tells you how much revenue it couldn't match.
It breaks quietly. A renamed property, an expired connection or a prompt someone "improved" fails without anyone noticing until the report looks odd.
Side by side
| | Make + ChatGPT | Rumorly | |---|---|---| | Reads free text | Yes | Yes | | Named source (person, podcast, AI assistant) | If you build it | Yes, with a dictionary | | Review queue for new names | No | Yes | | Overrides improve later answers | No | Yes | | Stripe MRR per channel | No | Yes | | Self-reported vs UTM / first touch | No | Yes | | Unmatched revenue shown | No | Yes | | Cost | Make plan + API usage + your time | From $49/month |
When to switch
When someone asks "how much revenue came from the founder's LinkedIn?" and the honest answer is "let me export two CSVs".
What we say about a Make + ChatGPT scenario comes from their public docs and pricing pages, checked on 22 October 2026. Products change; if something here is out of date, tell us and we'll fix it.