The Productivity Puzzle in 2025: Turning Point or False Start?

In the UK, the productivity conversation tends to follow a similar pattern. First we get a new data set that suggests some promise. Then there are some statements that highlight caution flags, and we either move on to a new conversation or simply redefine the conversation without asking the right and important questions. This year seems different - not because the headline numbers are strong all of a sudden pre-pandemic - but because there are multiple strands suddenly starting to converge. Statistically, the official estimates show output per hour running ahead of the pre-pandemic baseline. Survey data suggests less pressure on prices, with either steady demand or indicators of forward demand. Meanwhile, the conversation about AI adoption within firms is moving along and the conversation has transitioned from waiting to see to early mainstream adoption. None of these guarantees a step-change, but it signals that we might be narrowing the gap between hope and practice.

The article takes a measured, balanced view. It begins clarifying what the latest data is actually suggesting. It then reflects on where the modest improvements of 2025 might have come from and what levers leaders might choose to pull first across capital, skills, and management. The article concludes with a one-page checklist that managers can drop into the huddles of their teams, with the checklist also informing managers on how to communicate that information to teams in a way that's different than gathering as a team for a team huddle.

What do the latest numbers actually say?

Per the Office for National Statistics, output per hour in the April to June quarter of 2025 was approximately 1.5 percent above the 2019 average. So the UK is only very modestly ahead of its pre-COVID level. However, the same period shows a boost in output per hour and output compared to the previous year, as output per hour had dropped a year prior. The message is clear. The level effects are positive; the rate of change has not improved.

Key takeaway. Output per hour is above 2019; the annual comparisons have softened; thus, the question is more about momentum rather than level.

The leading indicators indicate there is a little light. The October flash PMI from S&P Global edged up, with respondents holding firmer confidence and further cooling in price pressures.. These are not GDP numbers and we should never over-interpret a single flash PMI number, but they suggest a better environment for firms to plan, invest and reorganise.

Key takeaway. The PMI flashes are early signals. A modest increase, and easing price pressures make it less costly to operationalise any changes.

Where did 2025’s modest progress come from?

The starting point should be investment, since capital deepening expands the potential effective capacity of every hour worked in the economy. Revised ONS estimates show that business investment fell by 1.1 percent in the second quarter of 2025 relative to the first quarter, which is a lower decline than the initial business investment reporting, which was essentially flat year-on-year. That quarterly wobble relates to timing and revisions to the sectors themselves, in particular Information Communications Technology equipment and other machinery. What can we be certain of? The investment line is fragile and the fundamental quality of projects selected and delivery becomes even more important.

Key takeaway. Investment is not falling off a cliff but fragile. Gains will depend on the quality of project selecting with a clear line of sight to throughput and unit costs.

The second ingredient is firm level reorganisation. The productivity improvements referred to by many of the leaders are not about using all brand new, shiny technology, but standardising routines, tightening measurement and shedding workflow issues that had built up during the pandemic. This is also the reason why the quality of management often acts as a multiplier to capital and skills.

Capital, skills or management: what should come first?

One convenient way to sidestep the debate is to ask: 'Where is an extra unit of effort most binding'? Studies across manufacturing and services types repeatedly show structured management practices say they can increase productivity by improving goal setting, tracking and continuous improvement of performance. Simply put, good management increases both machine and training return on investment. For growing firms, especially outside the frontier, this can be a very actionable starting point, since it turns diffuse change into regular habits.

Key takeaway. Management discipline is often the binding constraint; the mechanism that multiplies returns on capital expenditure and training by converting strategy into repeatable routines.

Is AI finally diffusing in a way that moves the needle?

Adoption is increasing from a low position. ONS business surveys show that nearly a quarter of UK firms were using some form of artificial intelligence by late September 2025, up from single digits when the question was first asked in 2023. Earlier ONS work also linked quality of management to higher rates of technology adoption. The implication is simple. AI will improve productivity when leadership links tools with process design, data cleanliness, and training, not when it is handed over to ad-hoc experimentation.

Key takeaway. AI adoption has now reached significant levels, but to realise benefits, leaders will need to address clear role definition, clean data and short feedback cycles.

What are the bottlenecks inside firms?

The typical challenges originate not from algorithms but from organizational culture.

• Data quality. Variations in definitions, redundancies and poor traceability will ruin any promising pilots and lead to a noisy dashboard.
• Workflow design. Often, tools bolt onto existing systems rather than integrating properly, forcing staff to double-enter information or bypass them entirely
• Capability gaps. In particular, supervisors, and certainly frontline teams do not get the 60 second or even 5 minute, job-specific training needed to make their habits easy to change.
• Incentives. For example, incentives are designed around activity, instead of throughput, defects, or customer outcomes.
• It's not glamorous work. More like a cadence of weekly decisions, visible metrics and quick documentation that demonstrates to staff what good looks like.

A one-page checklist managers can use next week

Teams adopt change more quickly when the message is simple and visual. Create a single annotated page and use text pictures to make the structure unmissable in shift briefings. A lightweight tool such as Adobe Express is perfectly adequate for laying out the checklist with clear headings and callouts.

Management checklist for a 4-week improvement cycle

• Target. Name one process, one metric and your desired weekly run-rate.
• Baseline. Capture today’s output per hour and unit cost, with a time-stamped snapshot.
• Change. Document one workflow change, one risk control and one short training item.
• Data. Define the input source, the owner and a simple error-logging rule.
• Review. Hold a 20-minute weekly huddle that inspects the metric and decides go, hold or pivot.
• Communication. Publish the before-and-after plots and a 100-word note to staff.
• Sustain. After week four, lock in the gain by updating standard work and coaching materials.

Key point. The checklist is a forcing function. It keeps change small, measured and visible, which is how teams bank gains rather than talk about them.

How should leaders measure progress without fooling themselves?

Anchor your measure of productivity in the cornerstone definition of productivity: output per hour. The OECD's standard allows you to make comparisons over time, and between countries, and it highlights what ultimately matters, which is the amount of value produced for each hour of labour. Use activity metrics only as leading indicators. If a dashboard is busy, but unit costs are not improving, then the system is not working yet.

To keep measurement clean, develop three simple rules.

• Measure output per hour and unit cost at the process level, adjusting for mix as relevant.
• Avoid double counting when digital tools change work from one team to another team.
• Incentivize based on customer value and quality, not the number of tasks clicked.

Key point. Output per hour and unit cost are your anchors. If they do not move, the productivity story is not real yet.

What would convince sceptics that 2025 is a turning point?

Three catalysts for the story to shift from false dawn to inflection.

• A sequence of improved prints. Several quarters where output per hour rises both on the quarter and on the year--rather than simply being measured against 2019 mark. That requires both macro stability and firm delivery.
• Broad-based adoption. ONS surveys which show that not only the volume of AI use is improving, but the sectors and firm sizes adopting while training and change management capacity are also being built.
• Higher quality investment. Not simply more investment, but a more considered project mix. The Q2 wobble highlighted why selection and delivery discipline is important.

Key takeaway. A real inflection will look like sustained quarterly and annual forward momentum alongside broader diffusion of good practice and a better investment mix.

Practical communication that accelerates adoption

When expectations are clear, behavior will change. Turn jumbled memos into a small set of short, contextual examples and place them where work actually occurs. If teams work off of screens, build a small library of examples for the new way to work. If the content is busy, either crop or simplify it instead of simply adding more decorations. Well labelled pictograms with small writing can align a diverse workforce in the same direction in less than a minute.

Key takeaway. Clear visuals and tight exemplars lessen cognitive load, thus reducing time to learn and increasing stability of the new routine.

Closing remarks

The United Kingdom is modestly above its pre-pandemic productivity baseline, but that momentum has not shamed the skeptics. The leading indicators indicate slightly more breathing room. AI adoption is approaching a point where it may matter in the aggregate, notwithstanding the ultimate benefits remaining local and contingent on pairing tools with clean data, structured management and wise investment. The most plausible path from here is not dramatic. It is a weekly rhythm, selecting one process, measuring what matters and communicating progress regularly. If your teams would rather have tangible clarity, circulate a one-page summary document and update the graphics and text-based pictures every Friday. The puzzle will not solve itself; steady routines can convert a fragile year into scaffolding for stronger and stronger ones.

FAQ

Is productivity in the UK finally above the pre-crisis trend? Yes. Output per hour in Q2 of 2025 is about 1.5 percent above the 2019 average, although lower than the previous year.

Do PMI surveys provide any indication of improvement? They suggest some mild improvement. The flash PMI for October edged up and the price pressures moderated; good news for planning and investment.

Is business investment contributing to productivity? Business investment remains fragile. Q2 2025 saw minor quarterly fall in business investment, nearly 0 year-on-year. Selection and discipline in delivery is paramount.

How prevalent is AI within British firms? Nearly one in four of companies reported some form of AI use by late September 2025, up from 9 percent in 2023. The advantage derived from AI is contingent upon management and quality and management of data.

What would be the cleanest way to measure productivity within a firm? Use output per hour and unit cost as anchors, as aspected to OECD definitions, and use activity dashboards as leading indicators only.