Learn and improve

What's working. What isn't. From your own data.

Conversation Intelligence shows booking rates, drop-off by question, and conversion totals — all from closed conversations, all without needing to export a CSV. Available to client admins directly from the operator platform.

See it in actionRead the docs
Conversation Intelligence metrics
  • Drop-off rate per question% of conversations that disengaged at each question
  • Booking rateconversations resulting in BOOKED as % of total
  • Conversation totalstotal conversations, total booked, total timed out
  • Minimum sample gatedata only shown when 20+ conversations completed
  • Self-serviceaccessible to client admins — no Friyay involvement needed
Drop-offshown per question — as a percentage of conversations that reached it
Booking rateprimary campaign performance metric — updated as conversations close
20 minimumconversations required before intelligence data is shown
Self-serviceclient admins access their own data — no export required
What conversation intelligence shows

The numbers you need to improve the campaign.

Conversation Intelligence surfaces three core metrics: drop-off rate per question (which questions lose contacts most frequently), booking rate (the primary success metric), and conversation totals. All three update as new conversations close — no manual refresh, no export required.

  • Drop-off rate per question — shown as % of conversations that reached that question
  • Questions listed in campaign sequence order — maps directly to campaign config
  • Booking rate — BOOKED conversations as % of total closed conversations
  • Conversation totals — total run, total booked, total timed out visible at a glance
  • Minimum 20 conversations required before data is shown — no premature conclusions
  • Client admins can access their own data directly — no Friyay involvement required
Intelligence data flow
  1. 1
    Conversation closes
    BOOKED, TIMED_OUT, or CLOSED state
  2. 2
    Analyser runs
    Drop-off point and conversion signals identified
  3. 3
    Pattern detector
    Booking rate and drop-off table computed
  4. 4
    Minimum sample check
    20+ conversations required before display
  5. 5
    Intelligence view
    Metrics available to client admin in operator platform
  6. 6
    Proposals generated
    Pattern detector creates proposals from findings
Using the data

From observation to action.

Conversation Intelligence is the starting point, not the endpoint. The drop-off table identifies where to investigate. The learning engine's proposals tell you what to try. Version history tracks whether it worked.

Drop-off table

The drop-off table shows each question in sequence order with its drop-off rate — the % of conversations that reached that question but did not answer it satisfactorily. High drop-off at Q2 usually means Q2 is the wrong question for where the contact is in the conversation.

How it works: Drop-off table tested with 25 tagged conversations — questions shown in correct sequence order with accurate rates.

Booking rate

The booking rate is the primary campaign performance metric. It shows BOOKED conversations as a percentage of all closed conversations for the campaign. This is the number that matters — and it updates every time a conversation closes.

How it works: Booking rate computation tested in the pattern detector suite — uses contributed_to_outcome tags from the analyser.

Self-service access

Client admins can access Conversation Intelligence from the operator platform without contacting Friyay. The data is scoped to their own campaigns — they cannot see other clients' data, and Friyay operators can see all clients.

How it works: Intelligence view requires manage_campaigns permission — client viewers are redirected. Client admins see only their own data.

Learning engine feed

Conversation Intelligence data feeds the pattern detector. When the drop-off table shows a persistent problem at Q3, the learning engine generates a proposal to address it — with the drop-off rate as the evidence and a projected improvement if the change is made.

How it works: Pattern detector generates proposals from the same signals that power the intelligence view — the data model is shared.

Want to see your campaign
intelligence dashboard?

We'll run the analysis against your closed conversations and show you the drop-off table.

Book a discovery callTalk to the team