Which Phase Is Your Outbound Actually In? A 3-Stage Test for Whether Your System Knows When to Reach Out

Probably phase 2, and that's worth checking rather than assuming. Most teams running email, LinkedIn, and a CRM alert or two call it "signal-based," but the phrase alone doesn't tell you how the system actually behaves. Here's a 3-phase test that does.

The framework below isn't mine. It's one I came across in the NextGen Sales Systems newsletter, and I couldn't independently re-confirm more than that this round, so treat the sourcing as a pointer, not an authority.

The 3 phases

Phase 1: one signal triggers one action. A form fill sends an email. A job change fires a LinkedIn connect. No filtering at all. Everyone's automation looks identical at this stage, because there's nothing underneath it beyond "event happened, message goes out."

Phase 2: one signal plus basic filtering. Before the action fires, the system checks whether the person is already in the CRM, on a do-not-contact list, or already a customer. Real progress over phase 1, but it's still reacting to one isolated event at a time. Most teams calling themselves signal-based are stuck here, not phase 3, whatever the pitch deck says.

Phase 3: signals get evaluated together before the system acts, not one at a time. Four things have to be true:

1. Identity resolution. A pricing-page visit, a LinkedIn comment, and an email reply three weeks later get treated as one person's behavior, not three unrelated events.
2. One timeline per identity across channels, not five separate dashboards checked by hand.
3. Signal scoring that decays with age. A reply this week outweighs a website visit four months ago.
4. A confidence threshold. The system only moves into automated action once accumulated signal weight actually clears a bar, not the moment any single signal fires.

Where my own system actually sits

I ran this test on trigger-finder, the signal system I built for my own outbound, before handing it to anyone else. It already does most of phase 3: identity-adjacent clustering ties related signals to one person, and a points model scores signal type so a reply and a page view don't count the same.

What it doesn't do yet: decay signal weight by age. A four-month-old website visit and a same-week email reply currently carry the same fixed score once they're both in the model. That's an open gap, not a solved problem, and it's more useful to say that plainly than to claim a polished system that isn't finished.

Run the test on your own outbound

Pull up whatever you're running, in-house or a vendor's pitch, and check it against the four phase-3 components above. Not which phase the sales page claims. Which phase it's actually operating in today.

If it fires on one signal with no filtering, that's phase 1. If it filters against a do-not-contact list but still reacts to isolated events, that's phase 2, and that's most systems on the market right now. If it resolves identity, holds one timeline, decays signal weight, and gates action on a confidence threshold, that's phase 3, and ask exactly which of the four components is missing, because something usually is.