List the processes your teams repeat each week, how often they happen and how long each one takes. The calculator turns that into hours and cost a year, and gives an indicative view of how much of that time AI could take on.
Use the average salary of the people doing these processes. On-costs cover employer NI and pension, usually 15 to 25% on top of salary. Cost per hour = (salary + on-costs) ÷ (hours a week × working weeks).
Count how often each one happens across the whole team, not per person.
Hours a year = times a week × minutes each ÷ 60 × working weeks. Cost = hours × cost per hour. One full-time equivalent is hours a week × working weeks.
ROI compares the value of the hours AI could take on with what you would spend on AI tools, setup and support in a year. ROI = (value − cost) ÷ cost. It assumes the freed time goes into more useful work, such as selling, rather than being cut.
Each type of work has a typical share that AI and automation take on in insurance broking and similar regulated operations, reduced if the process is already partly automated. These are indicative figures to help prioritise. Timing a process properly will give you the real number.
Any process that handles client personal data is flagged. These are often the biggest wins, but they need the right controls first (approved tools, a UK data region and anonymised data) before any AI is used on them.