See what your conversations reveal.

See what people need, how the conversation unfolds, and where your agent could help them more.

Explore individual conversations and the patterns they share.

One conversation · editing designs · enterprise, EMEA

Abracadabra identifies tasks and signals in the conversation. Every finding stays connected to its source. Illustrative example.

Understand one conversation.
Then see the pattern across many.

Move between a conversation, a recurring issue and the people affected by it.

See where the conversation changes.

Look for moments of confusion, repeated questions or changing sentiment. Open the relevant turn to understand what happened.

t1 · +0.6
t2
t3
t4
t5
t6
t7 · the turn
t8
t9 · −0.81

“Where are the rest? Can you try that again?”

Turn 9 · two turns after the agent sent 3 of 9 sizes.

  • scored on the person's turns, so the agent's tone never flatters the number
  • the turn where it changed is marked, so there is something to open
  • one sour conversation is noise. the same drop on the same intent, across hundreds, is a fix

Understand the steps within a task.

Follow the steps a user needs to complete. This example shows a resize that needs more help at the second step.

task · resize this banner for every channelIllustrative example · 9 turns
complete find the formats The 9 social formats are identified in the conversation.
failed resize it 6 of the 9 sizes come back blank.
not confirmed export the files Only the 3 sizes that rendered are exported.
Task status · unresolved. Review the user’s follow-up in session replay.

Follow a user’s experience across sessions.

Bring a person’s sessions together to see what they have tried and where they keep needing help. Open any point to review the conversation.

Session 1

Resize a logo · done

Session 3

find the formats · done

Session 5

Banner resize · unresolved

Session 7

Banner resize · unresolved

open replay

Session 9

Same task · failed again

Three attempts at the same task reveal a recurring issue worth investigating.

Compare the groups that matter to your team.

Use plan, region, returning-user status or your own metadata to build a group. Then explore its tasks and conversations.

  • saved once, then available to search, charts and monitors
  • open a cohort to reach its people and their conversations
  • behaviour works the same way: returning users is a field like any other

Know where to focus your attention.

Explore the topics and tasks people bring to your agent. Compare unresolved requests to find the areas that need a closer look.

36%
29%
22%
18%
14%
9%
editing
designs
templates
brand
assets
account
settings
integrations
billing

Task failure rate by topic · 12,400 conversations read, up 18% on the previous period · illustrative data

See who encounters a recurring issue.

Compare a signal across user groups to understand where it appears most often. Follow the evidence to see what each group needs.

Signal · Unresolved request · product-wide rate 11%

enterprise, EMEA +19
EMEA +12
returning users +3
enterprise −7
self-serve −9

Points above or below the product-wide rate · illustrative data

any signal task progress, tool events, user corrections, repeated intents, clarification failures, sentiment shifts, off-topic drift, escalation intent, and any custom event the agent already emits.
any slice plan, region, returning users, or any field already attached to the person. saved cohorts work here too.

Read a summary. Check the context.

Get a summary of the task, progress and signals in a conversation. Follow a highlighted finding back to the turn it describes.

  • written for a conversation, a person, a topic or a cohort
  • the status comes from the turns, not from a thumbs-down
  • nothing is claimed that has no turn under it

The conversation · Illustrative example · 9 turns

  1. t1Resize this banner for every channel we post on.
  2. t2Social only, or email headers too?
  3. t3Social only.
  4. t4Found 9 formats.
  5. t5Great, go ahead.
  6. t6Resizing for the 9 formats now.
  7. t76 of the 9 sizes came back blank. Sending the other 3.
  8. t8Here are 3 sizes, ready to post.
  9. t9Where are the rest? Can you try that again?

The summary

The user wanted the banner resized for every social channel. The scope was confirmed, but 6 of the 9 sizes came back blank and the agent sent the other 3 anyway. The user asked where the rest were.

  • Task statusfailed
  • Failing intentresize it
  • Decisive signalUnresolved request, turn 9

Find the steps that need more support.

In this example, 2,400 attempts led to 1,536 finished summaries. Explore the remaining attempts by step to see where users need help.

task · resize this banner for every channel 864 of 2,400 attempts stopped · illustrative data
  • resize it528 stopped · 22.9%Review these conversations to understand why resize it loses them.
  • export the files244 stopped · 13.7%The conversation ends without every file exported. Review the context to understand why.
  • find the formats92 stopped · 3.8%The user needs help pointing the agent at the right formats.
open the conversations 864 unresolved attempts · illustrative data

See what your conversations can tell you.

Explore user needs and recurring issues, with the conversations always close at hand.

Request access Open the playground

Explore the workspace with example data.