A warning about world growth becomes useful to a business only when it leads to a question the business can investigate. On April 10, 2024, Reuters reported an IMF assessment that global growth could fall to 2.8% by 2030 without substantial reform or technological progress. That conditional outlook is the starting point here, not a forecast for any individual company.
The practical question is narrower: what stops available people, equipment and knowledge from producing a reliable result? The discussion below develops an independent operational framework around an illustrative workshop. It does not describe a surveyed business, reproduce the IMF's model or estimate a national growth effect. Its purpose is to distinguish an improvement that merely moves work elsewhere from one that genuinely releases capacity.
Start with the result, not the busiest machine
Imagine a workshop that cuts components, finishes their surfaces, inspects them and packs accepted orders. Its manager notices that the cutting equipment sometimes stands idle. Buying faster equipment would make little sense if the existing machine already waits for work. But even keeping it continuously occupied would not necessarily help: additional components might simply accumulate before inspection. A local activity measure can improve while the customer receives nothing sooner.
The first task is therefore to define the useful output. For this workshop, a sensible starting unit is an accepted order that reaches dispatch with its required documentation. Counting unfinished components would answer a different question. Counting shipments without checking returns would also omit part of the result. The definition should remain stable during the comparison, otherwise an apparent productivity gain might come entirely from changing what counts as completion.
This does not make machine utilisation irrelevant. It changes its role from an objective into diagnostic evidence. Idle time at cutting, queues at finishing and repeated inspection all help explain the path of an order. None should be interpreted alone. A manager needs to know whether an idle resource lacks demand, lacks materials or waits for a downstream stage that cannot accept its work.
Make waiting visible before proposing automation
Follow one order from acceptance to dispatch and record when responsibility changes hands. The order may spend little time being worked on but a long time awaiting approval, a missing measurement or the next scheduled batch. Those intervals are not interchangeable. Waiting for a safety-critical inspection may be necessary; waiting because nobody knows which drawing is current is a different problem with a different remedy.
A useful record captures the reason for each delay without turning into a surveillance exercise. Staff should be able to distinguish missing information from equipment failure and competing priorities. If every pause is recorded as an operator problem, the resulting data will support the wrong intervention. The purpose is to find a repeatable obstruction in the process, not to assign blame to whoever happens to hold the order when it stops.
The record can stay simple: order identifier, arrival time, release time and the reason work could not proceed. Define whether overlapping delays are recorded separately or as one interval before adding them together. Otherwise the same waiting hour could appear under both a missing drawing and unavailable materials, making the total longer than the order's actual elapsed time.
The workshop can then ask whether automation removes an obstruction or merely accelerates the arrival of work at it. A faster cutting cycle cannot approve a drawing or resolve a disagreement about acceptance criteria. Conversely, an administrative improvement will not repair worn equipment. Observing the sequence first makes it easier to choose a change that addresses the actual cause rather than the most visible symptom.
A faster stage can create a slower system
Suppose cutting releases larger batches while finishing still handles orders in small groups. The cutting team may report fewer interruptions, yet finishing now has more material to sort and store. A rush order becomes harder to locate. A drawing revision affects more unfinished pieces. These are possible consequences of the illustrative arrangement, not measured outcomes or a claim that large batches are always undesirable.
Batch size should instead be tested against the whole order path. Smaller releases may reduce waiting but require more frequent preparation. Larger releases may economise on preparation while increasing unfinished stock. The comparison needs both sides. Choosing the option with the lowest preparation time alone would ignore the resources consumed by storage, handling and delayed completion.
Responsibility matters here. If each team is rewarded only for its own recorded output, it has a reason to pass more work forward even when the next team cannot use it. A shared completion measure creates a basis for coordination, but it does not erase the need for local measures. The useful combination shows both whether orders finish and where the process struggles to finish them.
Separate a genuine improvement from a changed mix
An order book does not necessarily contain the same work every week. A run of simple components may produce more completed units with unchanged methods. A smaller number of complicated orders may require more preparation and inspection while creating a more valuable result. Comparing raw counts without recognising that difference could reward an easier schedule and penalise demanding work.
The workshop need not build an elaborate model before learning anything. It can compare similar orders, identify unusual jobs and retain the original classification. If a new method is tried on only the easiest items, that limitation should be explicit. Extending the result to all work would require further observation, especially where tolerances, materials or customer requirements differ.
Prices introduce another distinction. Higher revenue per employee can arise because prices increased, not because more useful work was completed. Physical completion, realised revenue and resource use answer related but different questions. Keeping them separate allows a manager to see whether an improvement is operational, commercial or some combination of the two, without treating any one measure as a universal productivity score.
Count the work transferred outside the measured team
A process can look cheaper after tasks are moved to another department. If operators stop entering information and an administrator reconstructs it later, the production team's recorded hours fall. The organisation has not necessarily saved time. It may have added a handover and a new opportunity for error. The correct boundary includes the work required to make the same completed order usable.
The same issue arises with suppliers and customers. Asking a customer to correct every uploaded file may reduce internal processing time while making the service less convenient. Buying pre-finished components may remove a workshop stage but increase purchased input costs. Neither choice is inherently wrong. Both require an honest account of where the work and expense have moved.
A before-and-after review should therefore keep a short boundary statement. It identifies included staff time, purchased services, handling and corrective work. Where a cost cannot be measured reliably, the review should say so and test whether the conclusion depends on it. A visible limitation is more useful than a precise-looking saving calculated from only the favourable part of the process.
Quality is part of output, not an afterthought
Faster dispatch is not a clear improvement if more orders come back for repair. The workshop should retain a way to connect later corrective work with the original order. Otherwise a short trial may count the apparent benefit immediately and leave the cost outside its observation period. That timing problem is especially important when defects become visible only during customer use.
Not every return proves a production failure. A changed customer requirement, transport damage or an incomplete specification may require a different response. Categories should be narrow enough to guide action without becoming so detailed that staff cannot use them consistently. The aim is to understand why accepted work needed another pass, rather than to combine unrelated events into one reassuring average.
Safety and required checks remain constraints on any experiment. Removing a mandated inspection is not simply another way of saving labour. The trial must preserve the conditions under which the product is acceptable. If a proposed method changes those conditions, it needs a separate approval process before its speed can be considered a benefit.
Skills matter at the point where decisions change
Training attendance does not demonstrate that a new method works. In the illustrative workshop, a course on reading drawings would be useful only if staff can apply it to the decisions that previously interrupted orders. The follow-up should therefore examine actual handovers: are ambiguities recognised earlier, are the right questions asked, and can the next team proceed without reconstructing missing information?
Learning also needs a place in the schedule. If every employee is expected to maintain normal output while mastering a changed process, the organisation may hide the transition cost in overtime or unfinished work. Recording that cost does not discredit the change. It allows a fair comparison between the temporary effort of adoption and the recurring effort required after the method becomes familiar.
Transferability is a further test. A process that works only when its original designer is present has not yet become a dependable organisational capability. A second team should be able to follow the instructions, identify exceptions and know when to seek help. The documentation should capture the decisions that matter, not simply provide a longer description of routine actions.
Test a change without promising its result
A bounded trial begins with a specific proposition. For example, checking that drawings and acceptance criteria are complete before releasing an order may reduce repeated clarification during production. The workshop should identify which orders enter the trial, what remains unchanged and what would count as an adverse result. Those choices should be made before seeing the outcome.
A comparison period helps, but it is not automatically a controlled experiment. Demand, staff availability and equipment condition can change at the same time. The review should retain those differences rather than attribute every movement to the new checklist. Where possible, comparable orders handled under the previous method provide an additional reference, subject to the same quality and safety requirements.
A compact trial record
- Define the obstruction and the proposed change in plain language.
- Record which orders, teams and resource costs are included.
- Track completion, waiting, corrective work and exceptions together.
- State the conditions for extending, revising or stopping the trial.
- Keep unresolved measurement problems visible in the conclusion.
The decision at the end is not restricted to success or failure. A method might help one order type but create additional work for another. It might require a small design change before a broader test. Recording that narrower result protects the organisation from turning a promising observation into an unsupported universal rule.
Released time is an option, not automatic extra sales
Suppose the trial reduces the time needed to complete comparable orders without increasing returns or moving work elsewhere. The workshop has evidence of a useful operational change. It does not yet have evidence of additional demand. Freed capacity can remain unused if customers do not place more orders, if another stage remains constrained or if the required materials cannot be obtained.
There are several possible uses for the time: shorter lead times, maintenance, staff development or additional accepted work. Each has a different route to value and different evidence. A maintenance interval, for example, should not be presented as extra sales merely because it became possible after an efficiency improvement. The decision should be reported in its own terms.
This distinction also prevents double counting. If the same released hours are assigned simultaneously to more production, more training and reduced staffing, the plan promises incompatible uses of one resource. A capacity decision must specify which use has priority and what will be deferred if the available time proves smaller than expected.
What the wider growth debate can—and cannot—take from this
An individual process review cannot establish a national productivity trend. Firms differ, activity shifts between them, and improvements may coincide with changes in employment, demand and investment. Adding several workshop anecdotes would not solve that measurement problem. The macroeconomic outlook and the operational example belong at different levels of analysis.
What the example provides is a discipline for discussing possible improvements. Define the output, keep the resource boundary stable, investigate the obstruction and test the proposed change. Then ask whether the result survives quality checks, a different order mix and use by another team. These questions make a proposal more concrete without attaching an invented growth contribution to it.
The productivity gap is therefore not just a distance between two headline numbers. For a manager, it is also a set of unanswered questions about how work becomes an accepted result. Addressing those questions will not guarantee a particular economic trajectory. It can, however, distinguish a credible operational improvement from an attractive purchase, a displaced cost or a busier process that delivers no more.
Your perspective
Share your thoughts on the story.