Insurance agency service desk KPIs: what to measure and how

1. Requests per day, by type

Count what arrives, and split it into certificates of insurance (COIs), ID cards, endorsements, billing questions, renewals and everything else. The mix matters more than the total: a week heavy in endorsements is a heavier week than the count suggests. Calculate it as requests received divided by working days, per type.

2. Turnaround time, at the median and the slow end

For each type, note when the request arrived and when the customer got their answer. Report the median (the middle request) and the 90th percentile (the request that took longer than nine in ten). Track the slow end, not just the average: the requests that take three days are the ones customers remember.

3. On-time rate

Decide a due time for each type, then calculate requests completed by their due time divided by requests completed. A falling on-time rate is the earliest warning that the desk is overloaded, long before anyone says so. Pause the clock while a request waits on the customer or carrier, or the measure punishes the desk for other people's delays.

4. Backlog and ageing

At the same time each week, count open requests and how old they are: under a day, one to three days, over three days. A growing over-three-days bucket means work is being started and not finished, often because it is waiting on someone with no follow-up.

5. Rework rate

Count what came back: a certificate reissued, an endorsement corrected, a customer who had to ask twice. Divide by requests completed. Rework is the hidden cost of rushing, and it is where errors and errors-and-omissions (E&O) exposure live.

6. Where the time actually goes

For one ordinary week, ask the team to note roughly how long they spend reading and sorting, looking things up, typing, and chasing. Most desks find the judgment work is a small share, and the rest is searching and retyping. Repeat it once a quarter.

7. Work per person

Count open and completed requests per person each week. The aim is not to rank people; it is to see when one person is carrying far more than the others, often because requests are routed by habit to whoever answered last time. An uneven spread is an early sign of burnout and of requests sitting with someone who is on leave.

8. Customer chasers

Count follow-up emails and calls from customers asking where their request is. Each one is a request that was not visibly on track, and each one costs the desk time to answer. A falling chaser count is one of the clearest signs that turnaround and communication are improving.

Making the numbers reliable

Numbers are only useful if they are recorded the same way every time. Agree what counts as "done" for each request type: the certificate sent, the endorsement submitted, the billing question answered. Record arrival time as when the email arrived, not when someone opened it. If two people log differently, the KPIs measure the logging, not the desk.

Reporting to the owner

Owners need a short view, not a spreadsheet. A one-page weekly summary with requests by type, on-time rate, the over-three-days backlog and a single line on anything unusual is enough. Show this week against the previous one, so trends are visible without anyone having to read the detail. Keep the detail available for the service manager, who uses it to decide what to change.

Setting due times

On-time rate depends on sensible due times. Set them per request type, in business hours, from when the request arrives: for example same business day for certificates and ID cards received before mid-afternoon, two business days to submit an endorsement to the carrier, and longer for policy checking or audits. Write them down, tell the team, and review them after the first month of measurement. Due times you always miss are as unhelpful as none.

Putting them to use

Shlix IBS keeps these measures from the moment it starts reading the inbox: volume by type, turnaround, on-time and overdue work, and who is carrying what, downloadable as Excel. The first month replaces estimates with your own measurements.

• Review the numbers weekly, at the same time, with the people who do the work.
• Change one thing at a time and watch which number moves.
• Set a target only after a month of baseline data, and set it per request type.
• Put your own hourly cost against the time before deciding what to automate.

Common mistakes and what they cost

Each of these is a common route to an errors-and-omissions (E&O) problem. None of it is legal advice; follow your carriers' rules and your agency's procedures.

• Tracking only the average turnaround. A few very late requests disappear in the average, and those are the ones that become complaints and E&O claims.
• No due time per type. Without a target, there is no on-time rate and no early warning.
• Ignoring rework. Errors are fixed quietly and never reduced, and the agency has no record of how often certificates are reissued.
• Setting targets before measuring. Targets that are too easy hide problems; targets that are too hard are ignored.

Questions