Firstly, people were interested in where this kind of chat sits and how it compares to webchat. Consensus so far is that it’s best used for existing customers, but not only for easy conversations. Some ‘secondline support’ can work well.
Thinking of planning, there’s a morning peak caused by overnight chats, but demand is the same weekends and weekdays. Then we asked more.
How did you start? We already had successful, well-established, capacity-planned departments for webchat and social media. To gain further deflection of inbound calls, WhatsApp was selected as a next step in our digital journey. DLG engaged a group of consultants, for a 6-month test-and-learn pilot. When we had it working as desired, we needed to plan for it as BAU, like the other channels.
What was the challenge? Our big challenge, as planners, was the capacity plan and budget. At first, we thought Async was different. We asked colleagues in other organisations how they plan for this. Most staff to a fixed level, moving people on and off chat, as queues grow. This is not cost effective as we scale up on Async.
So, we stepped back and realised that it’s not really a different kind of forecast; we could approach it as we do for a new insurance product launch. We looked at what data we had, applied assumptions, and forecast accordingly.
Key decisions about data for models
- Do we forecast individual messages or unique conversations? In WhatsApp, you message when you have time and want to. So, messages are the demand we forecast, and this pattern is what we schedule to.
- Opening hours. Our chat window is open 24/7 but the team work core hours (like webchat). So we forecast hourly 24/7 but drop overnight chats into the first two hours of opening.
- What AHT? No systems give you average handle time in the same method as calls or back-office tasks. What we could get was ‘conversations closed per online hour’. From this we built a typical day’s work, removing shrinkage time to get to a productive hour. We translated this into seconds and divided by closed chats per hour. Now we have a working AHT assumption.
Completing the planning puzzle
When we had all the pieces of the puzzle, we used an erlang calculator to create the profile that our scheduling team can work to. Three key steps remained. Firstly, we can now measure our actual data to compare with forecasts, for efficiencies, service level, AHT etc. Then we could set useful performance targets for the operational team and compare across different teams or channels.
Finally, we can now analyse cost benefit by channel and challenge our strategy. Above all though, we feel confident we have staff in at the right times of the day and week. This is a good place to be as planners.
Author: Richard Carless and Colin Thornback at Direct Line Group
Date Published: 27/04/2021
This article was first published in the 2021 Best Practice Guide - Unlocking Opportunities: You are the Key
To download a full digital copy of the Best Practice Guide, click here