Value Stream Maps for Claims and KYC Operations

Value stream mapping is the Lean technique that looks simple in a textbook and is rarely done properly in regulated operations. Teams go through the motions. They draw the boxes, they add the times, they colour some waste. Then they present at the steering committee and nothing changes, because the map was not granular enough to drive action and not anchored enough to drive prioritisation.

This post covers value stream mapping specifically for claims operations and KYC operations, because those are two of the operational engines where the technique is most under-exploited and most valuable. The patterns we describe here are drawn from programmes in insurance, banking and wealth management, where claims and KYC are both high-volume and highly regulated.

Why VSM matters in regulated operations

Claims and KYC are high-volume transactional operations with specific regulatory windows. Claims have Consumer Duty response time expectations, complaint handling obligations, and policyholder outcome measurements. KYC has sanctions screening turnaround requirements, periodic review cycles, and customer experience pressure from onboarding drop-off. In both, the operational friction costs show up as customer outcome problems, regulatory findings, or both.

The VSM technique reveals where time actually goes. Most claims processes spend more than 80 percent of elapsed time in waiting. Most KYC journeys spend more than 70 percent of elapsed time in handoff or waiting for customer input. Those percentages are not anomalies; they are typical. The interesting question is where specifically the waiting happens and what can be removed.

What a good VSM contains

A VSM has more structure than a process map. At minimum, it captures:

Process steps

The sequence of activities, from trigger to completion. Each step is named, owned, and located (team and usually system).

Process times

How long each step actually takes when performed. Not the scheduled time, not the standard time: the observed time, ideally measured across a sample.

Lead times

The elapsed time from the end of one step to the start of the next, including waiting, handoff and queuing time. Lead times are where the big numbers hide.

Information flows

What triggers each step. What produces the output. Which system is updated. Information flows are often where rework and waste accumulate.

Material flows

The physical or digital artefacts that move through the value stream. In claims, this is the claim record, supporting documents, evidence. In KYC, this is the customer file, identification documents, screening results.

Customer flow

The customer's view. What they experience, what they wait for, what they hand over, what they receive. Customer-centric VSM makes the Consumer Duty conversation concrete.

Rolled-up metrics

At the end of the VSM, the total process time, total lead time, percent value-added, and the number of handoffs, queues, and rework loops. These numbers are the currency of the subsequent improvement conversation.

A claims VSM that reveals actual problems

Consider a motor claims process, from first notification of loss to settlement. A naive process map shows 12 steps. A value stream map of the same process reveals:

• First notification of loss intake: 8 minutes processing, then a queue of 6 hours 30 minutes before claim is opened in the system.
• Claim opening: 12 minutes processing, then 24 hours of waiting for document submission from the customer.
• Document review: 15 minutes processing (when the document is complete), 45 minutes (when it is not), and in 40 percent of cases the reviewer requests further documents, adding another 48 hours.
• Liability assessment: 25 minutes processing, but only starts when a random-delay batch job runs overnight, adding up to 22 hours of lead time.
• Repair estimate coordination: 35 minutes processing, plus an average 3 business days of lead time waiting for estimator availability and customer vehicle access.
• Settlement authorisation: 10 minutes processing, but passes through four approval tiers with average lead time per tier of 4 business hours.
• Payment: 6 minutes processing, executed in a daily payment run.

Total process time: about 2 hours of actual work. Total lead time: 9 business days on average, 14 days when documentation is incomplete on first submission. Value-added ratio: under 2 percent.

The map reveals where to look. The overnight batch job is a 22-hour queue invented by legacy IT scheduling. The tiered approvals add nothing that an automated decision table could not do for 90 percent of claims. The document back-and-forth is a symptom of unclear first-notification guidance. Each of these is a specific, actionable improvement, not an abstract "speed up claims".

For a deeper treatment of claims-specific AI patterns, see our post on AI in claims operations for insurance.

A KYC VSM that reveals drop-off causes

Consider a retail onboarding KYC process from application start to account activation:

• Application start: customer enters data on the digital journey, 4 minutes typical.
• Identity document upload: 6 minutes typical, 18 minutes for customers using mobile in low-bandwidth conditions. 12 percent of customers drop off here.
• Document verification (automated): 45 seconds processing, but 8 percent of cases flagged for manual review with a 4-hour queue.
• Sanctions and PEP screening: 30 seconds processing, 12 percent of cases generate potential matches that queue for compliance review, with an average 6-hour wait.
• Source of funds assessment: when required, average 3 business hours elapsed waiting for customer to provide additional documentation.
• Internal approval: for non-straight-through cases, average 1 business day waiting for approver availability.
• Account activation: 2 minutes processing, runs every 15 minutes.

Total process time: 8 to 15 minutes of actual work. Total lead time: 5 minutes for straight-through cases (about 60 percent of volume), up to 2.5 business days for the most complex cases.

The map reveals the drop-off problem is concentrated at document upload, which is a bandwidth and instruction design issue, and the complex-case drag is concentrated in the compliance review queue and the approval tier. Each of these has specific remediation paths that are measurable.

See our post on AI in AML and KYC compliance for treatment of KYC-specific AI enablement patterns.

How to conduct the mapping

Step 1: define the scope with a trigger and a completion

The VSM covers from trigger (what starts the value stream) to completion (what ends it). For claims, from first notification to payment. For KYC, from application start to account activation. Define these explicitly. VSMs that wander into upstream or downstream activities become unmanageable.

Step 2: walk the process

Do not map from documents. Map from observation. Walk the process with the people doing it, watch it happen, time the steps, note the handoffs. Observed process times are not documented process times. Observed lead times are always longer than anyone expects. This step is non-negotiable.

Step 3: sample across volumes

Claims and KYC processes have variability. A VSM based on one or two observations misrepresents reality. Sample across a reasonable range: different claim types, different customer segments, different times of day, different operators. The sample sizes do not need to be large to reveal the dominant patterns.

Step 4: capture the data layer

In both claims and KYC, the data layer is often a hidden source of waste. Double entry between systems, reconciliation between mismatched records, data lookups that require switching between three systems. Capture these in the information flow section of the map. Our post on action data layer architecture covers the data dimension of operational effectiveness in depth.

Step 5: identify the waste categories

Lean identifies eight waste categories (transport, inventory, motion, waiting, overprocessing, overproduction, defects, skills underuse). In regulated operations, the three most common in our experience are waiting, overprocessing (approvals and checks that add no value) and defects (rework from incomplete or incorrect input). Identify them explicitly in the map.

Step 6: quantify the prize

For each identified waste, estimate the operational cost and the customer-outcome cost. Operational cost is time and therefore money. Customer-outcome cost is drop-off, complaint rate, dissatisfaction. Both matter, and the business case for remediation depends on both.

Step 7: design the future state

The future state VSM is a target, not a plan. It shows what the process would look like after the identified waste is addressed. Critically, the future state should name the specific enablers required: automation of specific decisions, consolidation of handoffs, redesign of customer input, elimination of batch jobs. Without enablers, the future state is aspirational.

Step 8: plan the transition

The transition from current to future state is a programme in its own right. It uses the change impact assessment, the stakeholder register, and the RAID log disciplines covered elsewhere in this series.

Common failure modes

Failure: desk-based mapping

The VSM is drawn from process documentation without walking the process. The documented process matches the intent, not the reality. The map reflects intent, misses reality, and produces improvement recommendations that do not address the actual problems. Fix: always walk the process.

Failure: single-instance mapping

One claim is timed, one customer's KYC is traced. The resulting VSM is a snapshot, not a representative view. Fix: sample, even lightly. Five to ten observations per scenario reveal the variability that matters.

Failure: missing the customer view

The VSM is internal, operations-only. The customer experience is absent. Consumer Duty review asks about customer outcomes and the VSM has no answer. Fix: customer-facing swimlane is mandatory in regulated operations VSMs.

Failure: the untested future state

The future state VSM is a collection of aspirations. No specific enablers, no quantified improvement, no transition plan. Fix: every future-state change has a named enabler, a quantified improvement, and an owner for the transition.

Failure: the VSM as presentation artefact

The VSM is produced for a steering committee slide, then filed. Nothing changes. Fix: the VSM is an operational artefact tied to a remediation workstream. The workstream owner uses it. The operations leader updates it as changes land.

Connection to the broader stack

VSM connects to:

• Business capability maps: VSMs are within capabilities.
• Process decomposition levels: VSMs usually sit at L3 or L4.
• Service blueprints: service blueprints extend VSMs with the onstage-offstage-backstage dimension.
• NFRs: VSM findings often surface NFR gaps.

External references

The Lean Enterprise Institute is the canonical home of VSM as a discipline. The FCA Consumer Duty guidance sets the regulatory frame for customer-outcome improvements in UK regulated operations. The NAIC claims handling standards provide the US insurance perspective on claims operations standards.

The short version

Value stream mapping in regulated operations is the technique that reveals where the time and the customer-outcome costs actually live. Done with the discipline of walked observation, sampled variability, quantified waste and named enablers, the VSM is the bridge between operational reality and remediation programmes. Done cosmetically, it is a presentation exercise that changes nothing.

Our operational workflow automation service and process mapping and documentation service include value stream mapping as the diagnostic step in transformation engagements. For claims and KYC specifically, the relevant industry solution pages and banking solution pages show the transformation patterns.