DataIQ World Congress
Park Lane Hilton - 5 November 2025
(Check against delivery)
Building Trust to Enhance the Role of Data in the Economy
Good afternoon.
It is a real pleasure to speak to you today, at a time of extraordinary change in data’s role in the economy.
Here in the UK, nearly a quarter of businesses now report using some form of artificial intelligence, or AI, in their work. It was less than a tenth only two years ago.
AI is only as good as the data it learns from — and data is only as good as the trust behind it.
Investments in data are transforming what people are able to do and, increasingly, what machines can do on our behalf.
A range of international estimates suggest that collecting and curating of data might amount to 2-4% of overall spending in the economy. That is equivalent each year in the UK to £60-120bn.
The Office for National Statistics, where I work, sits at the heart of this system. In an age of algorithms, the truth still needs a reference point — that is the role of national statistics.
As many of you will know, the ONS has come under strain since the pandemic. The experience underlines how vital quality and trust are to the role of data in the economy.
My goal today is simple: to explain what happened and what we are doing to restore quality and rebuild trust.
The Office for National Statistics – its role and history
I’ll start by telling you a bit more about the ONS. Its roots go back over 200 years.
Today, we are an organisation of around 5,200 people and annual spend of around £450mn, so under half of one percent of the UK data economy.
The organisation was formed in 1996, when the central statistics office and the office of population censuses and surveys were brought together.
That Central Statistics Office was created in 1941 to meet the need for data to plan the wartime economy, bringing together data on national income and expenditure, production and trade, on the labour market and prices into a consistent system of national accounts.
Meanwhile, the Office of Population Censuses dates back over 200 years, delivering the first Census in 1801.
And you can still see those first principles of our founding organisations in what we do today.
Today, ONS data underpins decisions across the public sector, by businesses and people across the country.
Let me give the example of consumer price statistics.
Each month, we publish around 2,750 individual price points, describing the level of consumer prices in the UK. We estimate them by sampling over 180,000 prices across 150 locations and tapping into over 40 million prices directly on rents, used car prices and rail fares. Direct measurement will add 300 million prices when we bring in supermarket scanner data next year.
The price points evolve – for 2025, for example, the representative basket includes longstanding items like flour and teabags, but also video on demand subscriptions and VR headsets.
To form our sample we draw on our business register currently covering 3 million businesses across the country – soon to double to 6 million when we enhance it next year.
When we measure prices, we’re not just tracking costs — we’re capturing how the economy feels in real time.
Those price series are part of the economy. They are the operational target of the Bank of England. They are referenced in regulation of utility prices and rail fares. They are used in pay negotiations and to index pension and benefit payments. They index over £600bn of government debt.
And then there’s the Census.
Most of you will have played your part in completing your questionnaire in 2021 – thank you. The 2021 Census made greater use of government records than ever before and was digital first, with 22 million households in England and Wales responding online.
From that first census to the latest AI models, our purpose hasn’t changed — to give the nation a reliable picture of itself.
Recognising that the Government has commissioned the ONS to deliver a census in 2031, we currently have a public consultation in the field to inform the topics covered.
An unsustainable expansion in outputs following the pandemic with far too much broader ambition alongside it and too little response to feedback
When the pandemic hit, the country’s appetite for real-time data exploded.
The ONS rapidly expanded its work to track how people were coping, how businesses were adapting and how the virus was spreading.
I was at HM Treasury at the time and know first-hand just how indispensable those real-time indicators were for guiding decision making through the pandemic.
But the surge in the demand for real time outputs left a mark. The scale of outputs we were producing wasn’t sustainable. Global trends, accelerated by the pandemic, were pulling in the opposite direction – lowering response rates and driving higher costs for household surveys.
There was also a broader reach for ambition, with the launch of large and ambitious programmes: a new integrated data service for all to connect administrative data across government departments and a drive to do away with the need for a survey-based census by using administrative data. The drive for new missed an opportunity to deal with legacy systems underpinning statistics production.
By 2022, we were doing too much, too quickly and on too many fronts. Quality of the regular outputs suffered and the programmes failed to achieve their full ambition.
Sir Robert Devereux’s independent review, published this summer, called it what it was: a failure of leadership, including a culture that discouraged challenge. His review, and that by the Office for Statistics Regulation that preceded it, exposed a system that had lost its focus and damaged trust of the people it serves.
But it was not all doom and gloom. Sir Robert recognised the expertise and dedication of staff at the ONS and his confidence that with the changes in flight that the ONS could rebuild.
FOUNDATIONS FOR RECOVERY
That review was a turning point.
Alongside it, the ONS immediately launched recovery plans across our surveys and economic statistics.
Leadership has been reset at the top.
As well as overseeing those recovery plans, Emma Rourke, now Acting National Statistician, immediately launched work to redirect resources behind them. In August, Darren Tierney joined as our new Permanent Secretary, with a clear mission to restore our reputation for producing high-quality statistics and rebuild a stronger, more resilient organisation. Penny Young started in September as interim Chair, and has reset the strategy for the statistical system, based on values of openness, innovation, efficiency and rigour. I joined in August too.
Our first foundation is a more tightly focused mission that prioritises quality over quantity
In September, we agreed a new mission statement for the ONS:
To deliver trustworthy, independent, high-quality statistics that underpin the UK’s most critical economic and societal decisions and inform the public.
We have stopped or scaled back on analytical work that is not required to improve the quality of our statistics. We are focusing the integrated data service capabilities solely on projects to improve the quality of our statistics, starting with one on labour market statistics. Pete Benton is leading our new focus here.
Those decisions have allowed us to reallocate within the ONS to create 150 new roles, 100 of which will be filled this year, in support of the Economic Statistics Plan.
That builds on £26.4mn of resource that had already been made available across 2025/26 to improve our household and business surveys.
We will publish a proposal next week to go further and narrow the range of statistical outputs.
We will be replacing high frequency outputs with dashboards, consolidating across related publications, and engaging departments where there is overlap between the ONS’s published outputs and their own work. As part of the plan we will improve our publications, through a new ONS website that will be delivered in 2026.
Our second foundation is new structures to align the organisation and stakeholders
Focus alone is not enough. The real magic only comes when all hands pull in the same direction.
To that end, we have brought together all aspects of statistical production so that the whole end-to-end process – from social and business surveys, through to methods, processing, analysis, and publication – now sits under common line management, under me, as Director General for Surveys and Economic and Social Statistics.
We are also changing our governance. Good governance is not red tape — it is the architecture of effective delivery and of trust.
We are resetting enterprise-wide governance, giving all Directors an organisation-wide role in the recovery. There is plenty that is urgent here, but one area I would particularly call out is technology. Since I have joined I have been struck by the sheer number of different solutions for very similar problems. We will use our new fora to agree what the strategic solutions are and our plan to migrate to them quickly. To bring in fresh thinking we will soon be advertising for a new Director General for Digital Data and Technology.
We have also created an external steering group that includes the Bank of England, the Office of Budgetary Responsibility, HM Treasury and the Cabinet Office, to shape overall priorities for improving our economic statistics.
Users of statistics are a massive part of the recovery story at the ONS. It is only by having strong connections with them we can work to shared priorities informed by what matters most for the decisions that the statistics are there to inform.
Our third foundation is a set of principles to guide a virtuous cycle to restore trust
Structure and process can only get you so far.
The real test is cultural – rebuilding the way we think and behave under pressure.
We are embedding three principles to restore trust:
Creating Space for Excellence – so we set people up for success
Working in the Open – so we focus on what matters most
Leaders who support, encourage challenge and listen – so we can learn
Let me touch briefly on each.
First, Creating Space for Excellence
We are setting realistic ambitions that give people the space to succeed, develop and innovate. That means creating the right conditions — the time, tools and technology that enable people to do their best work.
We are carefully sequencing our improvement efforts to focus on what matters most. Taking on too much, on too many fronts, risks spreading capability too thin and setting teams up to fail.
Realistic ambition is not about doing less — it’s about building capacity to do things well. As capability strengthens, we’ll drive efficiencies that allow momentum to build naturally.
Alongside transformation programmes, we are embedding continuous improvement directly into production teams — giving them the bandwidth to solve problems at source and improve as they go.
Our new process-excellence framework guides how: reduce manual intervention, strengthening quality checks and cutting risk and duplication. It is being piloted in retail sales, drawing on lessons from a recent error.
The results are tangible:
- Of the 2.4 million business surveys we issue annually, 2.2 million are now online, with the remainder moving there shortly. That has already taken off £1.5mn out of postage costs alone.
- We are expanding use of administrative data and using it to replace surveys where that is possible. For the first time our migration estimates in November will no longer rely on the International Passenger Survey.
- Automation, including our bespoke ClassifAI tool, is already freeing hundreds of hours of work per year by automatically classifying products, occupations and sectors. That is set to increase to thousands as we go further.
Second, Working in the Open
This is simple, but powerful.
We’ve published our economic statistics and survey improvement plans and committed to regular progress updates.
Every quarter, we’ll release a public report showing exactly what we’ve fixed, what remains to be done and how we are measuring success. We will lay out future priorities and seek feedback on them.
We are backing that up with a new blog series where, when we reach a key milestone, we tell the story behind it and where we are going next.
We are also being open in our blogs when we uncover errors in our data. Alongside correcting the record, we explain what happened and the steps we are taking to prevent a recurrence. Our quarterly reports will review progress on these actions.
Third, Leaders who support, encourage challenge and listen
Having worked at the ONS for three months now I can only share how, like Robert Devereux, I have been impressed by the dedication and expertise of the people who work there. The sheer passion for the job and desire to improve statistics is quite something. With that, anything is possible.
For the record, given some of the reporting this week, that includes the many talented and dedicated colleagues I’ve met in several visits to the Newport office. The only complaint I have about Newport is it is sometimes hard to find a space in the canteen. I do not think one needs to look to a move that happened 17 years ago to explain why things went off track at the ONS over the last 3 to 5 years.
However, many people have in the past been treated badly and addressing that trust deficit is essential for moving forward.
It is why when Darren Tierney reset our mission, he also reset our approach to leadership. Each leader in the organisation is now signed up to a commitment “to build trust with our teams, to encourage challenge, to listen, and to take decisive action on our priorities.” We are embedding that into performance assessments, supported by 360 feedback.
The tone taken when errors occur is important. My starting point is there is likely to be a systems problem, which means the responsibility for what happened rests with leadership, ultimately me, rather than the team or individual involved.
And we are not just looking inward. We are talking to peers in sectors that have mastered learning from errors and near misses – aviation, nuclear, healthcare – to adapt their best practices for statistics.
To conclude, the Office for National Statistics is improving and has a bright future
It is early days, but three months into our leadership reset, there are already clear signs of recovery.
Foundations are moving into place.
The quality of our surveys has improved.
We’ve delivered important methods improvements to GDP data and, responding to feedback, improved the presentation of our monthly release.
Last month, we reinstated publication of producer prices, which had been suspended in April, when we uncovered a serious error. The reinstated data are better quality than prior to the occurrence of the error and our processes are stronger so there won’t be a repeat.
Working in the open, with these steps as clear proof points, is helping to rebuild trust.
The data community also has a special role to play in what happens next. Many of your companies will be feeding us data to power national statistics. Many of you will also be users of ONS data. Some of you may even be asked individually to respond to our surveys.
You will, I am sure, have ideas and contributions to make. Data are a public good and inherently a collective endeavour.
Get in touch, get involved and we will recover faster. We can build a stronger system and restore trust together.
Your National Statistics Institute Needs You.
Thank you for listening.