How to measure the performance of flow in service operations

1) What to measure (minimum viable set)

• Lead time (LT): calendar time from accepted to done.
• Touch time (TT): actual work time on the item.
• Flow efficiency (FE): FE = TT / LT (express as %).
• Work-in-progress (WIP): items currently in-flight.
• Throughput (TH): items finished per time unit (e.g., per week).
• First-time-right (FTR): % items completed without rework.
• Age of WIP: time already spent in progress for each open item.
• % Blocked time: proportion of LT where the item is blocked.

Optional but useful: queue length by step, handovers count, variability (σ, CV), SLA hit rate, arrival rate (λ).

2) How to collect the data (no new tool required)

• Add four fields to your ticket template (or spreadsheet): AcceptedAt, FirstWorkedAt, BlockedMinutes (sum), DoneAt.
• A simple rule: any pause > 15 min not under active work counts as waiting (not touch time).
• Mark blockers with a tag + reason; clear them with date/time to compute % blocked.
• Weekly export (CSV) of finished items; keep a second CSV for current WIP (with Age).

3) The few formulas you actually need

• Lead time: LT = DoneAt - AcceptedAt
• Touch time: TT = Σ active-work intervals (exclude waiting + blocked)
• Flow efficiency: FE = TT / LT
• Throughput (weekly): TH = #done in the week
• Little’s Law (sanity check): WIP ≈ TH × LT (use same time unit)
• FTR: FTR = (#done without rework) / (#done)

4) Weekly operating view (what to look at)

• Run chart of lead time (median + spread). Direction > precision.
• WIP vs Throughput (does WIP trend exceed delivery capacity?).
• Flow efficiency (is waiting the real constraint?).
• Age of WIP (which items are ageing towards breach?).
• FTR% (are we fixing quality or reworking it?).
• Top 3 blocker causes (owner + removal plan).

5) Example (claims team, weekly)

A claims team processes 120 items/week. Median LT = 10 days, median TT = 1.5 days → FE ≈ 15%. WIP averages 170 items; Little’s Law predicts ~TH × LT = 120 × (10/5) ≈ 240 item-days in the system; actual WIP suggests chronic overloading. After limiting WIP to 120 and enforcing a “ready” checklist, LT falls to 6 days, FE rises to 28%, FTR improves from 71% → 85%.

6) How to start in 60 minutes (field routine)

1. Pick one value stream. Agree the definition of Accepted and Done.
2. Add the four fields (AcceptedAt, FirstWorkedAt, BlockedMinutes, DoneAt) to your board/template.
3. Tag blockers with standard reasons (Access, Missing info, Waiting approval, Capacity).
4. Set WIP policy (visible limit + single intake policy per week).
5. Review weekly with the 6 charts/metrics above; decide one constraint to remove.

7) Common traps

• Measuring activity, not flow: counting tasks created or hours logged tells you nothing about delays.
• Blending “ideal” and “actual” steps: instrument the real path first.
• Overfitting KPIs: start with the minimum set; add only if a decision needs it.
• Punishing bad news: people will game metrics; use data to learn, not to blame.

8) Minimal data model (CSV headers)

ItemId, Title, Owner, AcceptedAt, FirstWorkedAt, DoneAt, BlockedMinutes, Rework (true/false), StepCount, Handovers, SLA_Target_Days

9) Takeaway

Flow performance is not a dashboard project. It’s a leadership habit: fewer items in flight, clearer entry rules, faster feedback on reality. Measure what shortens queues and reduces rework — everything else is decoration.