Let’s say you lead customer operations at a large insurer. You’re reviewing a proposal from your team to introduce an AI assistant into your contact centre. As you read through it, you see that your team estimates the assistant could improve productivity by 20%.
At first glance, that sounds promising, particularly if your team is handling more enquiries. The question is, could the assistant help them keep customer waiting times down without needing as many additional staff?
To explore that possibility, you’d first need to understand how much time the assistant could save on different enquiries. You could then estimate what those savings would mean for customer waiting times as volumes grow, with the same number of staff.
What does the 20% measure?
To see where the proposed assistant might help, let’s look at the kind of work your team does today. A typical enquiry might come from a customer with a question about their policy. To respond, a staff member may need to:
- Find the relevant policy details.
- Review the customer’s previous correspondence.
- Use that information to explain to the customer what happens next.
The proposed assistant would bring that information together and draft a response for your staff to review, helping them spend less time searching across systems and preparing an answer.
Before we can tell how much time that would save, we need to understand what the 20% estimate measures. I’d want to clarify both the improvement being claimed and the evidence behind it:
- Does 20% mean less time finding information, shorter calls, or more enquiries handled during a shift?
- Is that estimate based on a pilot, a supplier’s claim or an assumption?
To illustrate the difference, suppose an enquiry takes ten minutes to resolve, including two minutes spent finding information. Cutting that search time by 20% saves 24 seconds. If everything else stays the same, the whole enquiry takes 4% less time.
Across the enquiries your team handles each day, those seconds could add up. We’d want to understand whether that gives staff enough additional time to help waiting customers.
What could your team do with the time saved?
If your staff can finish those enquiries sooner, they could move on to customers waiting for help.
Whether that reduces customer waiting times depends partly on where the time is saved. An assistant that helps with complex policy questions may make little difference to straightforward requests. We’d want to understand how much of your team’s workload involves enquiries the assistant can help with, and whether it frees up time during busy periods.
With that information, your team could estimate how waiting times might change under different enquiry volumes and staffing levels.
The same reasoning applies when a proposal includes a cost saving. For example, reducing overtime would depend on freeing up time for the people and shifts currently requiring it. Time saved elsewhere may still be useful, but it won’t necessarily reduce that expense.
Once you’ve agreed which outcome to pursue, a business case can set out how it will be achieved and who will be accountable. In this example, you would own the service improvement. Your contact centre managers would help put the additional time to use, while your technology team would be responsible for the assistant’s performance and reliability.
Test the benefit before a wider rollout
With that outcome in mind, I’d compare a pilot with the current process using similar types of enquiry. The comparison would include the time staff spend reviewing and correcting the assistant’s work.
Then I’d look at whether the time saved helps your team deliver better service. To assess that, I’d use the pilot to answer a few questions:
- Can your team resolve more enquiries during a shift?
- Do customers spend less time waiting?
- Are enquiries being resolved properly, or do customers need to call again because an answer was incomplete?
Those results would give you evidence to assess alongside the costs of integration, training, ongoing operation and maintenance. Finance could help validate any claimed savings, while your technology team estimates the cost of running the assistant at the expected volume.
If the pilot shows that your team can resolve more enquiries while maintaining service quality, you can use those results to estimate the benefit and cost of extending the assistant to other teams handling similar work. As its use expands, tracking the results would show whether the expected benefits are being achieved.
When you review the proposal again, you’d have evidence to judge whether your team could serve more customers promptly at a cost that justifies the investment. You’d also know who would be accountable for delivering that result.