We can't prove what our AI returned Legacy eats the budget before we start Products take months to reach market
Value Buying workshop
The pain · unproven AI spend

Everyone is buying AI. Almost nobody can prove what it returned.

Around 70% of an insurer’s IT budget already goes to maintaining legacy systems. Before you approve another project, work out what it has to return to be worth it — and what evidence would actually settle it.

What would this AI have to prove?

1 · The ask 2 · Your context 3 · The test
Licences, integration, data work, internal time. The all-in number.
Add your context ↓
Total, run plus change.
Research puts this near 70%. West Monroe found 54% of carriers spend over half.
What any capital project must clear.
46% of carriers report 9–16 weeks. Rate indications alone run 16–30 days.
Whatever is on their slide. Put it in and test it.
$0 of verified annual benefit it must prove
Locked in legacy
$0
Discretionary capacity
$0
This project takes
0%

What the vendor has to put in front of you — in this order

1
A baseline agreed before anything was installed
Measured, not remembered. Without it there is nothing to compare against, and every later number is an assertion.
Required
2
A matched actual on the same measures
Same line items, counted the same way, after the fact. Different measures before and after is not evidence.
Required
3
Attribution you can defend
Why the result came from the system rather than headcount, market conditions, or a parallel project.
Defensible
4
Recovered capacity, in hours or FTEs
The number vendors rarely bring, because counting it needs your baseline. It is also the one that holds up.
Ask for it
—
“Up to X%” from someone else’s deployment
A reference customer’s best case with none of your baseline attached. Price it at zero.
Not evidence
This is a buyer’s tool, not a vendor’s. It does not tell you the AI will work. It tells you what it would have to deliver to be worth what you are about to spend, and what evidence would settle the question either way. If a vendor cannot meet the standard above, that is information — and it is far cheaper to learn it now than at the first renewal.

Why the money disappears before the AI arrives

The problem, its causes, and what each role is measured on — before anyone mentions a product.
The problem
We are under board pressure to deploy AI, spending real money on it, and we cannot show what it returned.
Causes
  • 74% of carriers still run legacy technology for pricing, rating and underwriting
  • Around 70% of the IT budget maintains those systems; 54% of carriers spend over half
  • 52% have delayed or cancelled two to three major technology projects under budget pressure
  • No baseline was captured before deployment, so nothing can be measured against afterwards
  • Vendor claims arrive as “up to X%” from somebody else’s environment
  • Benefits are projected at approval and never revisited at renewal
Measures — what each role is judged on
  • CIO — share of budget locked in run versus change, delivery against commitments
  • Chief Actuary / pricing — rate indication cycle, currently 16–30 days
  • Product / CMO — speed to market; 46% report 9–16 weeks for a minor update
  • CFO — verified return against hurdle, expense ratio, capacity redeployed

What proof looks like when it is done properly

Two documents from other sectors, shown because the standard travels. Ask any vendor for something that looks like these.
Insurance · value map

Accelerated Product Innovation

A carrier-facing value map built from the buying roles outward — recovered capacity upstream, faster time to market downstream, revenue as the consequence.

Read the case study →
Sample deliverable

Earned Value Report Card™

An independent assessment with a line-item baseline, the cost side included, and a matched actual at eighteen months. This is the standard.

See the specimen →

Value Buying for insurance

A working session for the people who approve the spend. How to set a baseline before you buy, write proof obligations into the contract, and test an AI business case the way your actuaries would test a rate filing. We do not sell insurance software — which is why carriers let us in the room.

Book a Value Buying session →