Market oversight: Data quality

High-quality member data is essential for pensions dashboards to function effectively. Without it, savers may struggle to locate all their pensions or understand the true value of their savings. We have an extensive programme of work in place to ensure schemes are preparing for their dashboards duties, with a heavy focus on data quality.

We first set out our expectations for ‘record-keeping’ (the management of member data) in 2010. This includes basic controls of:

• an annual review of ‘common’ data (personal data used to identify members), and ‘scheme-specific’ data (data relating to member’s participation in the scheme), to ensure they are both present and accurate
• schemes putting improvement plans in place where issues are identified

Since 2016, we have asked schemes to report their common and scheme-specific data scores to us in the scheme return.

We identified a risk that schemes without basic data controls in place would struggle more than others to comply with their dashboards duties. We want schemes to review and measure their data appropriately where they haven’t already done so, and put plans in place to improve it, so they are ready in good time for their upcoming dashboards duties. 

We launched a regulatory initiative last year, using the submitted data scores, to target the schemes that may not meet our expectations. Our main objective was to challenge these schemes to demonstrate to us how they met our expectations, and to encourage them to review and improve data quality. We also wanted to gain insight into how schemes manage data in practice and use this to refine our communications and guidance.

We’ve seen good progress in data cleansing, especially around dashboards preparation and common data, but value data assessments remain inconsistent and some schemes still pose risks despite overall improvements from the regulatory initiative.

Where we refer to trustees in this report, this can also include scheme managers where appropriate.

Published: 18 November 2025

On this page

Regulatory approach

Our initiative targeted schemes with upcoming dashboards duties. It ran from October 2024 to September 2025, with some follow-up continuing beyond this period. 

We mainly focused on common data as much of this data is required to successfully find dashboards users in their records. Schemes of all sizes were included in our approach, but we focused particularly on schemes with more than 1,000 members. However, the insights from this report apply to all schemes.

We sent schemes an email reminding them about our expectations around data. Then we took additional action if a scheme met one of the following risk indicators:

• They had never provided us with both common and scheme-specific data scores. These schemes were potentially at risk of not having looked at their data.
• They had not us with a common or scheme-specific data score that was measured within the last two years. These schemes were at risk of preparing for dashboards based on an out-of-date picture of their data.
• They reported a 100% complete and accurate data score, which is unusual and could indicate that their tests were not very robust. A sample of other schemes also received this intervention so we could understand measurement practice across the industry.
• They reported low scores relative to other schemes. These schemes likely needed to improve their data, and we wanted to confirm that they had a sufficient improvement plan in place.

Our interventions included:

• 2,411 emails reminding schemes about their duties in respect of measuring and improving data
• 847 targeted letters asking schemes to undertake a specific action
• 211 requests for information asking schemes to voluntarily provide more information to us about the actions they are taking to assess and, if necessary, improve their data

Insights from our engagement

Our review identified several recurring themes across schemes regarding data quality management and improvement efforts. 

Providing data scores to TPR 

Most of the schemes were able to provide us with their data scores when requested. For some, however, our engagement was the trigger for measuring their data. Other schemes told us they were undertaking extensive data review work (for example in preparation for buy-out), or were changing administrator, and were waiting for this to be complete before calculating and submitting their data score. 

Similarly, most schemes that had not provided us with a recent data score were able to do so in the 2025 scheme return.

Our general code requires trustees to assess the need for a data review exercise at least annually. Member data degrades and is subject to change. Therefore, most schemes will need to carry out a review at least once a year to make sure their data remains high quality.

Schemes must supply us with a data score if this is requested in the scheme return and is available.

Schemes are not consistent in defining scheme-specific data

For common data, almost all schemes we engaged with used the data items listed in our guidance.

For scheme-specific data, we acknowledge that the required data items will naturally vary depending on the benefits provided, member status, applicable legislation, and the strategic objectives of each scheme. However, there are several common elements that should be present across all schemes which we found were not consistently included in the schemes we engaged with.

At a minimum, scheme-specific data needs to cover the data items a scheme needs to calculate pension values, process member queries and events, and specific features of the scheme (such as contracting out or remediation exercises).

More information on data item identification is included in our updated scheme member data quality guidance.

Inconsistent data score assessment and calculation

Schemes typically used their administrator’s or administration software provider’s standard testing methodology to calculate their data scores. Both the presence and accuracy of the data items were generally tested. However, some of the accuracy tests were more simplistic than others.  

In some cases, schemes did not apply the scoring methodology appropriately. They were calculating the score based on the number of data items failed, and not based on the number of members for which one or more data items failed. In some cases, they calculated one score to cover both common and scheme-specific data. 

Additionally, some schemes with multiple sections only provided us with the score for one section. If a scheme calculates separate scores for different groups of members (eg for defined benefit and defined contribution members), it will then need to combine these to provide a single overall score for the scheme.

Trustees, or their administrator on their behalf, should assess member data in multiple dimensions to check its quality and identify if improvements are needed. Simply confirming that each field contains data is not enough.

Trustees need to submit separate scores for common and scheme-specific data.

The data score is the percentage of members in the scheme with a fully complete and accurate data record.

More information on data quality assessment and data score calculation is provided in our updated scheme member data quality guidance.

Varying levels of detail in data assessments

The level of detail in the data quality assessment reports varied and some reports lacked detail on the testing methodology used. 

We found trustees’ understanding of how data assessments have been carried out could be inhibited by how reports were formatted and where detail surrounding testing methodology was missing.

For example, where the standard tests or information are not relevant to the circumstances of the scheme, this is not always clear and could be confusing to the trustees.

We did, however, see some good practice where the administrator has provided the trustees with additional explanations where needed, and whether any actions are required.

Data reports should provide sufficient contextual information to enable the trustees to understand which data items are being tested, how they are being tested and the impact of identified issues on the scheme.

More information on data reporting is provided in our updated scheme member data quality guidance. 

Further work needed on improvement plans

We observed widespread data cleansing and improvement activity. However, improvement plans tended to be informal or fragmented, and some schemes only had vague timescales (eg ‘by Q2 2026’). Very few schemes provided us with standalone documents that showed they had followed all the steps in our guidance.  

Where improvement is required, there should be a clear record of the plan so that trustees can refer to it easily. A standalone document means the trustees don’t have to locate multiple documents or items of correspondence to monitor progress.

An improvement plan should include objectives, the scope of work to be undertaken, a clear timeline, resources, and clear outcomes. It should also set out the roles and responsibilities of the trustees and administrator, identify dependencies with other work and explain how improved data will feed back into the scheme.

The trustees should work with the administrator to ensure that data improvement actions are appropriately prioritised for the needs of the scheme and the resources available.

More information on improvement plans is available in our updated scheme member data quality guidance.

Data improvement focused on matching requirements

Most dashboards improvement efforts focused on preparing the data needed to find savers in their records. Few schemes were actively preparing the pension value information needed for member communications. This ‘value’ data will be a key focus for us in our next phase of industry engagement on dashboards data. 

Trustees need to understand what data they need to return to members and by when. They should assess the quality of this data and put plans in place to ensure they can meet their dashboard duties.

More information is available in our dashboards guidance.

Not all trustees took active roles in data management 

Many trustees did not initially involve themselves in the responses to us, instead deferring to administrators. The reliance on administrators was a recurring theme throughout the intervention. For example, administrators typically defined the ‘scheme-specific’ data items to be measured, with little challenge from the trustees, and not all trustees scrutinised the data reports they received.   

However, we have also noted some examples of good data governance, including quarterly discussions on data at trustee boards, and sub-committees focused specifically on administration, including data.

Trustees should take an active role in the management of member data. They remain accountable even if administrators do the day-to-day work.

Trustees should ensure administrators have adequate controls and processes in place to maintain good quality data, obtain regular reports and scrutinise these to be satisfied they provide a true assessment of the scheme’s data quality.

You can find out more about roles and responsibilities around data quality in our updated scheme member data quality guidance.

Next steps for data quality improvement

Pensions dashboards are the catalyst for many schemes to focus on their member data. Trustees must ensure that data improvement work is delivered so that they are able to meet their dashboards duties, and this needs to cover both the data used to find savers in records (‘match’ data) and the information they need to return to savers (‘value’ data).

However, high quality member data is not only relevant to pensions dashboards. High quality data is needed to ensure that:

• scheme members can receive the right information and the right benefits at the right time
• trustees can minimise errors and mistakes, which can have serious negative impacts on members and be costly for trustees to resolve
• defined contribution schemes can process core financial transactions promptly and accurately
• all schemes can meet regulatory data requirements, including those introduced by pensions dashboards
• schemes can deploy advancements in technology and member communications to drive efficiencies and improve members’ accessibility and inclusivity
• members can engage with their pensions easily and effectively
• schemes can achieve their desired strategic ambitions such as buy-out or consolidation

Managing data quality is not a one-off exercise. We urge trustees to build on the momentum created by dashboards and to treat member data as the strategic asset it is. We have updated our guidance on data quality following the learnings from this intervention. It includes practical examples on how trustees can ensure that their scheme maintains good quality member data on an ongoing basis.  

In July 2025, we launched the next phase of our work on data quality, with a deep dive into the data quality controls and dashboards data preparations of the largest schemes in the UK. Further work looking at the data preparations of the wider universe of schemes scope for dashboards is being defined, for launch in 2026. Where schemes are unable to demonstrate how they meet our expectations, we may intervene. may include seeking voluntary improvements within a given timescale and/ or issuing improvement notices where appropriate requiring schemes to take specific action within specific timeframes or face enforcement action.