FR

Release flow expertise · GO FAST method

Reduce your batch release cycle time

Batches waiting without anyone knowing why, month-ends in waves, priorities that change every day: your release process is a black box. We make it visible, manage it and remove its bottlenecks one by one, with an approach inspired by the theory of constraints.

Results measured on our assignments

Trucks of medicines rolling again on a clear road46 → 34 daysaverage release lead time, in one month
Released pallets leaving the warehouse€6Mof stock released in six months
An investigator lifts the warning signs that were blocking the batches−49 %overdue anomalies blocking release
A quality technician calmly reviews a batch record2 days → 3 hfor record review by QA technicians

Results measured on several production sites.

Do you recognise your release process?

  • Teams spending their time chasing batches and replanning.
  • A month-end phenomenon, with waves of releases.
  • Release as a black box: nobody knows where batches are, or what is blocking them.
  • Vague release prerequisites, with a missing one discovered too late.
  • Priorities that change every day, in favour of the batches that make the most noise.
  • Management that takes up the issue when inventory is already rising.

Understand your flow to control it

1Robustness before performance

An average release lead time hides what matters. Two sites can show the same average and live opposite realities: one releases predictably, the other alternates between express batches and forgotten ones, with the chaos of priorities and the month-end waves that go with it.

Average lead time or box plotTwo sites have the same average release lead time of 35 days. The box plot reveals that site A ranges from 12 to 75 days, while site B stays between 28 and 44 days. The goal is to squash the box, then bring it down.0 d20 d40 d60 d80 dWhat the average shows35 d35 dSite ASite BTwo identical sites?0 d20 d40 d60 d80 dWhat the box plot showsSite ASite BSame average, two realities0 d20 d40 d60 d80 dWhat we’re aiming forBeforeAfterSquash the box, then bring it down
The average

A single figure that hides the spread: impossible to know whether batches go out on time.

The box plot

The box contains half of the batches, the whiskers show the extremes. The taller it is, the more unpredictable the flow.

The goal

First squash the box to make the flow predictable, then bring it down to gain days, for good.

2FIFO leads to OTIF, not the other way round

Pushing late batches or batches a customer is waiting for to the front seems logical. But every queue-jump overtakes others, which become tomorrow’s emergencies. Handling batches in order of arrival, with controlled work in progress, makes lead time predictable instead: commitments are met, and OTIF follows.

Prioritising by urgency or FIFOOn the left, batches overtake each other according to emergencies and one batch is forgotten: release lead times spread from 8 to 76 days. On the right, with FIFO, batches leave in order: lead times stay between 27 and 42 days.Prioritising by urgencyinout1forgotten5826374Batches overtake each other, some are forgottenBatch release lead time0 d20 d40 d60 d80 dPrioritising with FIFOinout87654321Each batch leaves in order of arrivalBatch release lead time0 d20 d40 d60 d80 d
The vicious circle of urgency and the virtuous chain of FIFOPrioritising by OTIF creates a vicious circle: a late batch becomes urgent, other batches are overtaken, and they fall behind in turn. Prioritising with FIFO and controlled work in progress makes lead time predictable, allows reliable commitments, and OTIF follows.Prioritising by OTIF: the vicious circleA batch is lateIt becomesurgentOther batchesare overtakenThey fallbehindThe chaosof prioritiesPrioritising with FIFO: the virtuous chainFIFO + controlled WIPA predictable release lead timeReliable commitments to customersOTIF, as a consequence
Controlled work in progress

FIFO alone isn’t enough: if the queue grows, each batch waits longer, even in order. We limit work in progress and plan at the pace of the bottleneck.

An express lane, strictly limited

Real emergencies exist, such as a risk of stock-out. They go through a dedicated lane, limited to one or two batches and decided in a ritual, not ad hoc.

OTIF, an outcome indicator

OTIF is tracked at management level. Using it to reprioritise batches day to day is what creates chaos.

3The bottleneck moves: every step must be managed

Each day, your site releases no more batches than its slowest step. But a release flow chains steps that are sometimes synchronised, sometimes in parallel, and the bottleneck moves over the weeks. Managing the overall cycle time isn’t enough: you need to see, at each step, what comes in, what waits and what goes out.

A step in the release flow shown as a pipeA step has an inflow, an outflow, a width that corresponds to the possible work in progress and a length that corresponds to the throughput time. It can be running normally, a bottleneck when work in progress builds up, or dry when it is waiting for batches.A step in the flow is a pipeInflowOutflowWidth = possible WIPLength = throughput timeThree possible situationsNormal operationAs many batches leaveas come in.BottleneckMore batches come in than leave:work in progress builds up.DryThe step is waiting for batches:its capacity is lost.
A bottleneck that moves from one week to the nextThe flow chains production, batch record review, quality control, deviations and QA review. In week 1 the bottleneck is record review, in week 2 quality control, in week 3 deviations. The site’s throughput is always that of the current bottleneck.The site’s throughput is that of its bottleneck… and the bottleneck movesWeek 1ProductionBR review · bottleneckQuality controlDeviationsQA reviewWeek 2ProductionBR reviewQuality control · bottleneckDeviationsQA reviewWeek 3ProductionBR reviewQuality controlDeviations · bottleneckQA reviewWithout managing each step, you improve in the wrong place: that’s the “whirlpool” effect.
What we measure at each step

Inflow, work in progress and how it evolves, actual throughput time against target, available capacity and versatility.

Why centralised management

To see where the bottleneck is today, anticipate tomorrow’s, and avoid local optimums that penalise the overall flow.

What this changes

Resources go where they really unblock batches, instead of speeding up steps that don’t limit release.

Our approach: manage the flow before trying to improve it

Even before speeding up a step, effective flow management already delivers results on the overall time.

Put out the fire, if needed
1

Put out the fire, if needed

When the situation has deteriorated, the Effective Quality Brigade runs a time-limited task force: batch piles sorted by age, normal flow isolated, target set and workload quantified.

task forceEffective Quality Brigade

Map the flow and its prerequisites
2

Map the flow and its prerequisites

With the qualified persons, we describe all the release prerequisites and the steps leading to them, as well as the data available, or not, in your systems.

release prerequisitesmapping

Measure accurately
3

Measure accurately

Work in progress, throughput and lead times at each step, analysed statistically rather than as averages: robustness comes before performance.

work in progress and throughputvariability

Manage by colours
4

Manage by colours

Centralised, visual management, run through short rituals, which shows where each batch stands and what is blocking it. It delivers results even before the steps are improved.

visual managementrituals

Plan around the bottleneck
5

Plan around the bottleneck

The counter-intuitive but decisive approach: plan all activity at the pace of the slowest step. The leadership team is on board from the start.

theory of constraintsleadership team

Remove bottlenecks one by one
6

Remove bottlenecks one by one

Identify the bottleneck, plan around it, improve it, then start again with the next one: batch record review, deviations, quality control or organisation.

continuous improvementbottlenecks

Three projects, three results

Injectables production unit

Twelve days of release time gained in one month

An average release lead time of around 46 days and finished-goods inventory to reduce. The batch record review flow was identified as the priority lever: review rules, on-line review and management.

  • Records reviewed on the line, during production
  • Two review steps maximum before release
  • Daily management of the review flow
46 → 34 daysaverage release lead time, in one month
CDMO site, sterile dosage forms

Millions of euros of stock released

Finished batches blocked by unhandled anomalies and finished-goods inventory rising sharply. Management of anomalies and high-stakes batches, a targeted task force and investigation coaching.

  • Open anomalies: 212 → 117
  • Overdue anomalies: 119 → 61
  • A product quality support department created afterwards
€6Mof stock released in six months
CDMO site, second entity of the same group

Record review that no longer takes days

The same approach applied to a second entity, focused on the quality assurance record review flow.

  • Record review redesigned with the QA technicians
  • Release flow management
2 days → 3 hfor record review by QA technicians

See all our case studies

How an engagement works

Map
1Map

Release prerequisites with the qualified persons, detailed steps, matrix of available data. The leadership team is introduced to the theory of constraints.

Measure and manage
2Measure and manage

Introduction for managers, data collection tools, visual management rolled out in the departments and observed on the shop floor until it runs on its own.

Remove bottlenecks
3Remove bottlenecks

A first improvement loop on the main bottleneck, with our methods for batch records, deviations or organisation.

Option · Put out the fire first
+Option · Put out the fire first

A time-limited task force from the Effective Quality Brigade, to get back to a calm situation before improving the flow for good.

with the Effective Quality Brigade

Frequently asked questions about the release flow

What is release cycle time?

It’s the time between the end of a batch’s production and its release. It includes all prerequisites: batch records, quality control analyses and data, deviations, OOS, environmental monitoring. It is what weighs on your inventory and service level.

Why not manage cycle time by its average?

Because the average hides variability, which creates the chaos of priorities and month-end waves. We measure cycle time statistically, with work in progress and throughput at each step: that’s what lets you act in the right place.

What is the theory of constraints applied to release?

A release flow chains several steps, and the site releases no more batches per day than its slowest step, the bottleneck. The approach consists of identifying this bottleneck, planning all activity at its pace, then improving it, before starting again with the next one.

What are the most common bottlenecks?

Most often: batch record review, deviations and anomalies, quality control and the organisation around the flow. The bottleneck can also move over time, hence the importance of continuous management.

Should you start with a task force?

Only if the situation has deteriorated. A time-limited task force from the Effective Quality Brigade puts out the fire, by isolating the normal flow and handling the oldest batch piles. But it isn’t sustainable: you then need to manage and improve the flow.

Do you need software to manage release?

Not to get started: we begin with simple tools, fed by your data extracts. Our PERISQOPE tool can then centralise the data from your systems and replace scattered Excel files.

Are your batches waiting to be released?

Tell us about your situation in thirty minutes. We’ll tell you where your bottleneck probably is.

Tell us about your situation