AI in customer service: a business case without the hype
How to calculate AI in customer service: cost per ticket, where the saving is, which costs get forgotten and what to promise management. With an example.
Enter your own figures. The calculator works with cost per resolved ticket and includes what is often forgotten: your own time for knowledge work and the months before full effect.
The ROI calculator works out what AI in customer service could be worth to you from your own figures: ticket volume, staffing and cost, the share that self-service resolves and the time draft answers save. It shows time freed, the annual saving, the net result in the first year and the payback period. The model is the same as in our business case guide, and the result is an estimate based on your assumptions, not a promise.
An estimate based on your assumptions, not a promise. The saving is the value of time freed; it becomes money when the time is used for something else.
The model is cost per resolved ticket, as in our business case guide. It values the tickets self-service takes over and the minutes draft answers save, and then deducts what is otherwise forgotten.
The saving is the value of the time freed. It only becomes money when the time is used for something else: longer opening hours, a shorter queue, knowledge work or not having to bring in extra staff at peak season. Do not count on reducing staff in the first year.
Read four measures before the start and then every month: tickets per hundred orders, the share of draft answers sent unchanged, handling time per ticket and customer satisfaction per channel. If the measures do not move, the calculation does not hold, whatever it showed.
It starts from the cost per ticket: customer service's monthly cost divided by the number of tickets. Tickets resolved by self-service are valued at that cost, and the minutes draft answers save are valued at the hourly cost. Supportifier's annual cost, any one-off cost and your own time for knowledge work are then deducted from the saving.
From the worked example in our guide to the business case for AI in customer service: 1,300 tickets a month, two agents, a quarter of tickets to self-service, draft answers on half the rest saving three minutes each, three months before full effect and one day a week for knowledge work. Replace them with your own.
Expect the first months to cost more than they return. The knowledge base has to be written, agents review draft answers and self-service needs time to take over. The calculator accounts for that by counting no effect during the months you enter.
Not in the first year. The saving in the calculator is the value of the time freed, and it only becomes money when the time is used for something else: longer opening hours, a shorter queue or not having to bring in extra staff at peak season. If a position becomes redundant over time, it usually happens through natural attrition.
Tell us which question takes time today. We go through what a first step could look like.
See how the knowledge work, the review and the first channel fit together.
How the rollout works