The two-stream model

Resonant IQ sits on top of the tools you already use and reconciles them into a single read on each customer. To do that, it sorts every piece of evidence into one of two streams. Like the rest of the product, the streams name phenomena, not apps — so the model is identical whether your conversations live in Zendesk, Intercom, or Help Scout.

The two streams

A common trap is to think of account activity as “what your team does.” It isn't. It captures the customer's non-dialog moves too — a form they filled out, a meeting they took, a task they completed — as well as the times they didn't do something.

Whose move was it?

Within account activity, an internal flag records who acted — your team or the customer. That one bit changes the meaning of everything above it. “No contact in three weeks” reads very differently depending on which side went quiet.

Silence is evidence

Most tools only react to things that happen. Resonant IQ also reads what doesn't. Its detectors watch for the absence of customer activity — a champion who used to reply within the hour and has now gone dark, a scheduled check-in that never got booked, a form that stopped coming in.

That absence is what powers signals like a key contact going silent. You can't spot a gap by reading a single conversation; you can only see it against the account's whole two-stream history.

Why split them at all

Keeping the streams distinct lets Resonant IQ ask the questions that actually predict what happens next: Is the customer's tone in conversations drifting away from the cadence of activity around the account? Have we gone quiet on them, or have they gone quiet on us? Is the enthusiasm in a call backed up by anyone actually doing anything?

Every score, risk, and signal is built by reading both streams together. To see what those signals look like once assembled, read Signal & card types, or step back to How Resonant IQ thinks.