The metric
The Communication Risk Index
a number for the conversations in between.
Every regulated firm has a number for satisfaction and a number for complaints. Neither one moves until it is too late to act. This is the third number.
What is the Communication Risk Index?
The Communication Risk Index (CRI) is a 0–100 measure of conduct risk carried inside customer conversations, calculated by EchoDepth across every interaction rather than a sampled subset. It is built from congruence: the measurable gap between what a customer says and what the involuntary signals in their face, voice and language indicate they mean. A high score means the words and the signals agree. A low score means they diverge, which is the condition that precedes complaints, escalation and attrition. The Index is reported at three altitudes — one interaction, one function over thirty days, and one figure for the board pack. It differs from satisfaction and complaints metrics in direction rather than degree: satisfaction reports how a customer felt afterwards, complaints report what has already gone wrong, and the Index moves while the interaction is still in progress and the outcome can still be changed.
One score, three altitudes
The same measurement answers three different people, and each one can act on it.
Altitude 1
One interaction
Face scored across 44 FACS Action Units. Voice scored on pitch, pause and strain. Text scored on clarity, hedging and commitment. The three are fused into a single congruence score for that conversation.
Acts: a live prompt to the agent, during the call.
Altitude 2
One function, 30 days
Every interaction in a function rolled into a distribution. The bottom decile is where complaints originate, which tells a quality assurance team where to spend its time instead of sampling at random.
Acts: coaching and script changes aimed at one decile.
Altitude 3
The board pack
One figure alongside satisfaction and complaints. Satisfaction is usually stable. Complaints are usually stable. The Index is the one that moves first, and it moves in the direction the other two will follow.
Acts: a number that moves before the complaint does.
How it differs from satisfaction and complaints
| Measure | What it tells you | When |
|---|---|---|
| Satisfaction NPS, CSAT |
How a customer said they felt once it was over, from those who chose to answer. | After |
| Complaints volume, uphold rate |
What has already gone wrong, typically weeks after the interaction that caused it. | Too late |
| Communication Risk Index | Where the words and the signals diverge, across 100% of interactions, with an auditable record for each. | During |
A firm running all three has a leading indicator for the first time. Satisfaction and complaints both describe a decision the customer has already made. The Index describes the conversation they are still having.
What the Index is not
It is not a rating of an employee. The Index scores customer interactions, not agent performance. The same congruence signal read from the agent’s side is a separate measure with a separate governance basis, and it is not what this number reports.
It is not a decision. Every score is an advisory signal for qualified human review under Article 22 UK GDPR. No vulnerability classification, suitability assessment or escalation is made automatically.
It is not sentiment analysis. Sentiment scores the words. The Index scores the distance between the words and everything else, which is why a call can read as positive on a transcript and still score low.
It is not a single-channel score. Congruence only exists when face, voice and text are measured together. Where a channel is absent, the Index runs on the channels present and reports its confidence alongside the score.