CARTO's signal engine. Built by RevPack, run by CARTO.
RevPack engineered a signal engine in CARTO's Salesforce: public buying signals in, rep-approved outreach out. CARTO's GTM Ops team now runs it in-house.
.png)
Executive summary
What is a Signal Engine
Intent says something happened. A signal engine says what to do.
A system RevPack builds inside your CRM. It captures buying signals from the sources that matter in your market, qualifies them, and turns each one into outreach a rep approves.
A signal, not a score
Who acted, on what, and when. An intent score keeps one number.
One record per signal
Stored in the CRM with history and status, traceable to pipeline.
A human gate
AI drafts the outreach. A rep approves every send.
Before and after
The challenge
Before: alerts
- Alerts expired in a vendor UI. The intent platform flagged surging accounts with no signal behind them, no history and no attribution.
- Research was manual. For every message, a rep found the contact, read the account and wrote the email from scratch.
- Public buying moments never reached Salesforce. Reactions to CARTO posts, comments on competitor posts and GIS job openings were not stored anywhere.
The solution
After: a system
- Every qualified signal is a Salesforce record. Permanent, reportable and traceable to the sequence it started.
- Outreach is drafted before the rep opens the record. Built from the signal and the prospect's stack. The rep approves every send.
- Public buying moments are captured. LinkedIn engagement and GIS job openings pass the gates and land in Salesforce as signals.
How it runs
Six gates. One human approval.
Listen
Gate the post
Gate the person and company
Log it in Salesforce
Draft the sequence
Rep approves
“Building this on our own gave us the flexibility to make sure it was absolutely applicable to our organization. We feel empowered to run this and expand from here.”
“The speed and detail of their delivery were most impressive.”
FAQ
Questions about the engine. Answered from the build.
Does the system send outreach automatically?
No. A rep edits or approves every drafted sequence in Salesforce before Apollo enrolls the prospect from the rep's own mailbox.
How are false positives handled?
Most never reach a rep: the post gate and the person and company gate filter them out first. Signals a rep rejects are marked Rejected, and those decisions tune the gates.
What happens when a new signal appears during an active sequence?
Steps that have already been sent stay locked. Only the unsent steps are rewritten around the new signal, so the rest of the sequence reflects the latest buying moment.
What happens when enrichment providers do not have enough data?
Commercial providers are weaker outside North America and on small firms, so coverage varies by geography and company size. CARTO's engine records those gaps instead of hiding them.
Can the same architecture work with HubSpot or another CRM?
Yes. The gates and the signal ledger are the core; the CRM and the sender can change. On HubSpot the Signal Log is a custom object with native workflows and reporting, which is usually the fastest install. CARTO's implementation runs on Salesforce and Apollo.