Pick the wrap-up code for every contact from your own list
Picks the wrap-up code from your list when a contact closes and says if it was avoidable, for the agent to confirm. The codes are an example to replace.
Try it on this example
Channel (call transcript, chat, email): call transcript
The whole conversation
- Which wrap-up code fits the main reason the customer made this contact?Bill question100%
- Does the customer raise a second, different reason for contact in the same conversation?Yes97%
- Could the firm have avoided this contact?Our error or delay100%
- What underlying cause of this contact does the conversation establish?Communication failure99%
- Which service is this contact mainly about?Home broadband100%
- How does the contact end for the main reason the customer got in touch?Resolved in the contact100%
- Does the customer say they have contacted the firm before about this same matter?No91%
- Does the customer say the website, app, bill or letters were missing, unclear or wrong about this matter?Yes97%
- Does the customer name or clearly describe a competing provider or its offer?Yes98%
- Is there enough in the conversation to tell why the customer made contact?Yes99%
These are real answers stored from one run on this example.
The prism behind it
Pick the wrap-up code for every contact from your own list
Fields
- Channel (call transcript, chat, email)
- The whole conversation
Context
We are Veloway, a US home broadband and TV provider that also sells mobile plans to its broadband customers. The text is one whole customer contact that has just ended: a call transcript or chat with speaker labels, or an email thread. The answers fill in the wrap-up code the agent would otherwise pick, and feed the weekly contact-reason and avoidable-contact reports. They describe the contact, never the agent. How we code a contact: - One code per contact: the reason the customer got in touch, as they describe it, not whatever came up later in the conversation. A second, different reason is recorded as a second code by the agent. - A contact is coded by its topic, not by the mood of the customer. Complaints are recorded separately. - A contact is avoidable when it only happened because of something we did wrong or late, because the customer is chasing something we already promised, or because the answer is already on our website or in our app. A new, genuine need is not avoidable. - The root cause is what the conversation itself establishes. We do not guess at causes the conversation does not show. Worked examples: - "My internet drops every evening" is Fault or outage, even if the agent then books a visit. - "Why is my bill $20 more this month?" is Bill question, even if the agent offers a new plan. - "Where is the engineer? He was due at nine" is Visit or delivery, and it is avoidable as chasing. - "How do I change my Wi-Fi password?" is How-to question, and it is avoidable when the app or help center already explains it.
Questions
Which wrap-up code fits the main reason the customer made this contact? Choice
Pick the code for the reason the customer got in touch, using the definitions below and the coding rules in the context. When the customer raises several things, pick the one they contacted us about first, unless they say another matters more.
Does the customer raise a second, different reason for contact in the same conversation? Yes / No
Count a second matter only when it would take a different wrap-up code from the list and the customer asks for something about it. A follow-up question about the same matter is not a second reason. Yes: The customer raises a second matter that needs a different code. No: Everything the customer raises belongs to the same code.
Could the firm have avoided this contact? Choice
Use the definition of avoidable in the context, and judge only from what the conversation says. When more than one option fits, pick the first that applies in this order: Our error or delay, Chasing, Self-service would do.
What underlying cause of this contact does the conversation establish? Choice
Pick the cause the conversation itself shows, from what the agent finds or the customer explains. If the conversation does not establish a cause, choose the insufficient information option rather than guessing.
Which service is this contact mainly about? Choice
Pick the service the main reason for contact concerns.
How does the contact end for the main reason the customer got in touch? Choice
Judge the main reason for contact only, from the last exchanges about it. A second matter raised later does not change this answer.
Does the customer say they have contacted the firm before about this same matter? Yes / No
Count an earlier call, chat, email, ticket, visit or reference number about the same matter, however the customer puts it ("this is the third time I have called"). Do not count earlier messages in this same conversation, or earlier contacts about other matters. Yes: The customer says they contacted the firm about this matter before. No: The customer mentions no earlier contact about this matter.
Does the customer say the website, app, bill or letters were missing, unclear or wrong about this matter? Yes / No
Count only what the customer says about our own information: they looked and could not find it, it was unclear, or it was wrong. Do not count the agent's view. Yes: The customer says our information on this was missing, unclear or wrong. No: The customer says nothing of the kind.
Does the customer name or clearly describe a competing provider or its offer? Yes / No
Count a named provider or a clear description of one ("the other company on my street offers..."). Do not count a general wish to pay less. Yes: A competing provider or its offer is named or clearly described. No: No competitor is named or described.
Is there enough in the conversation to tell why the customer made contact? Yes / No
Answer No when the text is cut off, too short, mostly silence or hold music, or so damaged by transcription errors that a careful reader could not tell why the customer got in touch. Yes: A careful reader could tell why the customer made contact. No: The text is too incomplete or garbled to tell.
Lens columns
reason, reason_probability, second_reason, second_reason_probability, avoidable, avoidable_probability, root_cause, root_cause_probability, product_area, product_area_probability, outcome, outcome_probability, repeat_contact, repeat_contact_probability, self_service_gap, self_service_gap_probability, competitor_named, competitor_named_probability, enough_information, enough_information_probability
Run it on your own text
Add this prism in the app, change any question, and test it on a file of your own.