What your claims inbox tells you that your team cannot

Nobody lied

This is the part worth being clear about, because it sounds like an accusation and is not one. The estimate came from how busy the team felt, and the team genuinely was busy.

They were working a smaller number of much harder claims than anyone realised. Volume was low; difficulty was high; the felt experience was identical. Every person in that business would have given you the same wrong number in good faith.

That is the normal condition of an operation nobody has instrumented, and it is why we measure before quoting anything.

The four numbers worth having

• Actual count of claims received per month, from the inbox or portal rather than from memory
• Time per claim, from arrival to decision, including the time it sits waiting
• The exception rate — what proportion do not follow the standard path
• Where time is spent inside a claim, split between retrieval and judgement

The third and fourth are the ones that change decisions. A process where 15% of items consume 60% of the effort is a different problem from one where everything takes the same time, and it wants a different system.

Why the finding made the project smaller

A lower true volume meant a smaller system was appropriate. It reduced what was worth building and it reduced what could honestly be charged, because our fee is a proportion of a measured manual cost.

That is an uncomfortable incentive to sit with, and it is the reason the measurement has to happen before anything is scoped, agreed in writing, with the figure visible to both sides. A supplier who measures after quoting has every reason to find a large number.

The same engagement produced a second finding of the same kind: a part where the price a handler would reasonably have accepted was seven times a defensible market figure. Both are in the claims engine write-up, which also states plainly that the system was designed and validated but never built.

How to do it without a supplier

You do not need anyone to run this. For one month:

• Count every claim that arrives. A shared spreadsheet is enough.
• Note the timestamp it arrived and the timestamp a decision was issued.
• Tag each one as standard or exception, and why.
• Ask handlers to record, roughly, how long they spent looking things up versus deciding.

Four weeks of that will tell you more about your operation than any proposal will. If the number turns out to be small, you have saved yourself a build. The costing method turns the result into an annual figure.

What to do with an uncomfortable number

Sometimes the measurement shows the process is cheap and the felt pressure is coming from somewhere else — a reporting requirement, a difficult account, one product line generating disproportionate disputes.

That is a genuinely good outcome. Automating a process that was never the problem is the most expensive mistake available, and it is the one an unmeasured project makes.