Ask an operations leader how long it takes to settle a straightforward property claim or quote a mid-market commercial risk, and you’ll usually get an answer in days. Ask how much of that time someone is actually working on it, and the room goes quiet.
That gap between elapsed time and working time is where workflow friction lives. It’s the claim waiting in a queue for a document that was already received but filed in the wrong place. It’s the submission passed between three people because nobody owns clearance. It’s the adjuster rekeying an estimate from a PDF into the claims system, or the underwriter re-requesting loss runs because the first set was incomplete.
Friction rarely appears as a line in the budget. It shows up indirectly: longer cycle times, overtime, backlogs during busy periods, brokers placing business elsewhere, claimants calling to ask what’s happening. This post sets out a practical way to make it visible and put a number on it, using your own data rather than industry averages.
The five kinds of friction
Most operational friction in claims and underwriting falls into five categories. Naming them helps, because each one is measured differently.
- Waiting. Work sits idle in a queue, an inbox, or a pending status. Nobody is doing anything wrong, but nothing is happening either.
- Handoffs. Work moves between people, teams, or systems. Each transfer adds delay and a chance of something being lost or misunderstood.
- Rework. Work goes backwards: a file reopened, a quote revised because of missing data, a payment reissued, a document requested twice.
- Rekeying. People copy information from one place to another, such as email to system, PDF to spreadsheet, or one system to another.
- Searching. People hunt for information that exists somewhere: the latest estimate, the policy wording, the broker’s original email.
Step 1: Pick one workflow and define its start and end
Don’t try to measure the whole operation at once. Choose a single, high-volume workflow with a clear start and end, for example:
- FNOL received → first payment issued (for a defined claim type)
- Broker submission received → quote issued (for one line of business)
- Endorsement request received → endorsement issued
- Claim document received → document reviewed and actioned
Define exactly which events mark the start and end, and which cases are in scope.
Step 2: Reconstruct the timeline from data you already have
Most insurers have more process data than they realize. Claims and policy systems log status changes. Email systems timestamp messages. Document management systems record uploads. Task and workflow tools record assignments. Telephony records calls.
Pull those events together for a sample of cases and reconstruct each case’s timeline: every status change, assignment, document arrival, outbound request, and inbound call. Process-mining tools can automate this, but a spreadsheet and a few hundred cases will already tell you a lot.
From each timeline, calculate:
- Elapsed time: start to end.
- Wait time: periods where the case sat in a status or queue with no activity.
- Handoff count: how many times ownership changed.
- Loop count: how many times the case moved backwards, such as reopened, re-requested, or revised.
Step 3: Measure touch time directly
System logs show when things happened, but not how long people spent working. To estimate touch time, combine two lightweight methods:
- Work sampling. For a couple of weeks, ask a group of handlers or underwriters to record what they’re working on at random intervals, or to log time against a short list of activity types (reviewing, rekeying, searching, chasing, deciding).
- Shadowing. Sit with a few people for a few hours each and time the steps on real cases.
You’re looking for the ratio between touch time and elapsed time, and for how touch time splits between judgment (assessing coverage, pricing a risk, negotiating a settlement) and friction (rekeying, searching, chasing).
Step 4: Put a value on it
Now translate friction into money and outcomes, using your own figures:
- Capacity cost. Hours spent on rekeying, searching, and chasing each month, multiplied by a fully loaded hourly cost. This is capacity you’re paying for that isn’t going into judgment work.
- Rework cost. The number of loops multiplied by the average touch time per loop.
- Cycle-time cost. For claims, longer cycle times tend to increase handling expense and claimant contact, and can affect outcomes in lines where delay has a cost. For underwriting, slow quotes can lose broker flow. Estimate the value with your own data on lost quotes or complaint volumes.
- Contact cost. Inbound “where is it?” calls and emails generated by waiting, multiplied by your cost per contact.
Keep each assumption written down and agreed with finance. A conservative estimate that finance accepts is worth far more than a dramatic one nobody believes.
Illustrative example: a TPA analyzing its property claims might find that most elapsed time sits in two waiting states (awaiting documents and awaiting a reviewer) and that a large share of adjuster touch time goes on finding and rekeying information from incoming documents. Neither problem calls for a new claims system. Both are friction that intake automation and better routing can address.
Step 5: Find the root causes
Each kind of friction points to a different fix:
- Waiting often comes from batch processing, unclear ownership, or reviewers who are overloaded at certain times of day. Fixes include routing rules, workload balancing, and automated reminders.
- Handoffs come from specialization and organizational boundaries. Fixes include fewer, clearer handoffs, with complete context passed along each time.
- Rework usually traces back to incomplete information at intake. Fixes include completeness checks at the start and specific requests for what’s missing.
- Rekeying comes from unconnected systems and document-heavy inputs. Fixes include document extraction and integration.
- Searching comes from scattered information. Fixes include a single file view with evidence linked to source documents.
Step 6: Make friction a standing metric
Measurement shouldn’t be a one-off project. Once you’ve built the timelines for one workflow, keep them running and report a few friction indicators alongside your usual productivity and quality metrics:
- Touch-to-elapsed ratio
- Average wait time by queue or status
- Handoffs per case
- Loop rate: the percentage of cases with at least one backward step
- Share of touch time spent on rekeying, searching, and chasing
When you introduce automation, these are the indicators that show whether it’s actually removing friction or just moving it somewhere else.
Why this matters before you invest in AI
AI can remove a great deal of friction, especially rekeying, searching, and the rework caused by incomplete intake. But investments land best when they target measured friction rather than assumed pain. A friction baseline tells you where to start, gives you a credible before-and-after comparison, and keeps the program honest about what changed.
Find out where friction is costing you most. Take the Aketud Insurance AI Readiness Assessment to baseline one claims or underwriting workflow, identify the biggest sources of waiting, rework, and rekeying, and see where AI would make a measurable difference.
