The Order-Entry Cycle Time Nobody Owns End to End

This is a classic case of functional sub-optimization with no end-to-end process owner. Each function measures a local slice and is rewarded for local performance, but no one is accountable for the handoffs - and the handoffs are where the variance lives.

Why the local metrics look good:

• Sales 'order accuracy' is measured against what sales entered, not against what the customer needed. An order can be 98% 'accurate' and still be wrong for the customer.
• Credit '4-hour average approval' is an average. The 95th percentile is likely far worse; complex or exception orders may sit for days. Averages hide the tail that drives customer complaints.
• Warehouse '99% pick accuracy' says nothing about how long an order waits in the queue before it is picked.
• Billing '2-day invoice lag' is downstream of ship and does not affect the customer's delivery experience.

Measurement design flaws:

1. Local metrics with no roll-up to the customer metric. Nothing ties the function measures to order-to-ship cycle time.
2. Averages instead of distributions. The complaint is about unpredictability (the tail), so 95th percentile and standard deviation matter more than the mean.
3. No handoff measures. Time-in-queue between functions is unmeasured and therefore invisible.
4. No process owner to own the end-to-end metric, so no one is accountable for the interactions between functions.

Fix: appoint an order-to-ship process owner, define the end-to-end metric with a distribution (median and 95th), and add handoff/queue-time measures at each interface.

Use a three-tier hierarchy that connects the customer metric to controllable process measures.

Tier 1 - Customer/end-to-end (the 'why')

• Order-to-ship cycle time: median and 95th percentile, by order type.
• On-time-in-full (OTIF) against promise date.
• Customer-reported 'predictability' score (survey item).

Tier 2 - Process/handoff (the 'where')

• Queue time at each interface: order-capture→credit, credit→warehouse release, warehouse release→pick start, pick complete→ship.
• % of orders requiring exception handling (credit holds, address issues, partial stock) - exceptions are usually the tail.
• Rework loops: count and time of orders returned from credit to sales, from warehouse to sales.

Tier 3 - Functional (the 'who')

• Keep the existing functional measures but re-anchor them: credit approval measured at median and 95th percentile; sales accuracy measured against customer-confirmed requirements, not internal entry.

Governance: the process owner reviews Tier 1 monthly and Tier 2 weekly; function heads own Tier 3 but are accountable for their contribution to Tier 2. The key design rule: every Tier 3 measure must have a demonstrated link to a Tier 2 measure, or it is decoration. This ensures local success cannot coexist with end-to-end failure without being visible.

Acknowledge the legitimate concern - functional accountability for safety, cost, and quality should not be diluted - then separate the two accountabilities rather than trading them off.

Response structure:

1. Distinguish process accountability from functional accountability. The process owner is accountable for the flow and handoffs (cycle time, predictability, exception handling). The warehouse VP remains accountable for warehouse safety, cost per unit, and pick accuracy. These are different jobs, not competing ones.
2. Show the cost of the status quo. Every function is at target while the customer metric worsens and the 95th percentile is deteriorating. That is the definition of an unowned process. The warehouse VP's own metrics are being damaged by upstream delays (orders arriving late and in bursts), so the current arrangement is not protecting the warehouse either.
3. Offer a bounded mandate. Propose a 90-day trial with a written charter: the process owner can set handoff standards and escalate, but cannot change functional targets or headcount. Review results at 90 days.
4. Make the incentives consistent. Add a shared metric (OTIF or order-to-ship 95th percentile) to all four function heads' scorecards so the end-to-end outcome is in everyone's interest, not just the owner's.
5. Escalate only if needed. If resistance persists, the sponsor (the VP above all four) must decide; a process owner without sponsor backing will be ignored at the first conflict.

The principle: an end-to-end metric without an end-to-end owner is a number nobody can act on.