This research does not establish a reliable UK HVAC-wide customer acceptance rate. Test your customers’ experience instead of accepting either “everyone prefers AI” or “everyone will leave”.
Our recommended pilot measures are:
- Abandonment before useful details are captured.
- Repeated attempts about the same issue.
- Requests for a person and successful handovers.
- Incorrectly repeated details.
- Complaints attributable to the interaction.
- The proportion of calls that staff must redo.
Use a transparent opening that identifies the business and the automated assistant. Provide a clear route for callers who need a person or cannot use the digital process. Where calls are recorded, include the appropriate recording transparency described in Question 20. (ico.org.uk)
Test regional accents, interruptions, background noise and callers who cannot receive an SMS. These are proposed test conditions, not claims that a particular product fails them.
A useful launch boundary is one narrowly defined call category or one overflow period. Expand only when the office can explain the mistakes, the handover works and the overall customer outcome is acceptable.
Sources cited on this page
- ico.org.uk ↗ico.org.uk