Enhanced Identify & Verify
Successfully matching every caller and chatter to a customer record in our system to enable self-service, personalisation, and accurate routing that leads to first-time resolution.
Initiative overview
Background
A large retail energy supplier, taking over 200,000 voice calls and 60,000 chats every week. Starting with a traditional IVR ID&V solution that’s only successful 64% of the time, with no automated ID&V in chat. Accurately identifying customers is a regulatory requirement under the UK Data Protection Act.
Problem
The ID&V journey is a traditional IVR application with a success rate of only 64%, and is only available on little over half of inbound calls—and unavailable in other channels. If the ID&V isn’t completed here, the customer cannot use self-service bots and human agents are required to complete this manually with the customer, extending call duration.
Solution
There were a lots of different improvement areas for this, I generated around 20 user stories that we deliver over a couple of months.
- Wider deployment: make ID&V available on all appropriate phone lines and chat channels
- More options: extend the customer data items that could be used for identification/verification
- Better data: improve quality of data held in our CRM system for better matches by capturing missing details
- Make it easier: Implement Express ID&V journey for frequent contactors
Design approach
This journey was already completed successfully 99% of the time by human agents handling calls and chats, so we had a wealth of expertise in our contact centres to draw upon. Our compliance teams also maintained signed-off business processes for this area.
Channel specific considerations
This journey was upgraded to work with all customer channels, and we had to carve out quite a few special cases due to the fundamental differences between digital and voice channels.
- With chat we have no Caller ID but we do have web context, e.g. are they already logged in to the website? Is there a session cookie from the last chat?
- Some data items are pretty much impossible to capture reliably in voice, such as email address
- How do we make an address picker journey work in voice? With chat it’s an easy, selectable list. With voice, we had to go with a paged menu if the caller only gave a post code
- With chat it’s easy to use full name when it’s typed in, with voice capturing a customer or business name is less reliable
I also had to consider where it was not appropriate to offer ID&V. For example, our sales lines: these callers weren’t customers just yet! Also on certain sensitive complaints lines, or lines provided for third-party contacts who won’t have an account with us (e.g. other energy suppliers.)
Benefits
- Success rate of ID&V increased from 64% to 92% in voice and 98% in chat
- 1,250 additional agent hours saved each week, as they didn’t have to manually complete the ID&V process
- 1,200 new valid customer phone numbers / email addresses captured each week where we previously didn’t hold one
If I had more time
Exploring some more advanced options I’d seen other organisations use, such as multi-modal authentication via mobile app, or biometric voiceprint authentication.