Gears Co Working 1920x1080

Strategies and Resources for Optimising Pensions Dashboards Programme Matching Criteria

Data matching is one of the biggest PDP readiness hurdles

PDP ’s new staging guidelines are in place. The first schemes will be connecting by April 2025 and all schemes must be connected by Oct. 31, 2026. As a result, companies are moving quickly to either begin, re-engage or continue to advance their PDP projects to meet their staging dates.

One of the fundamental challenges that must be addressed to ensure successful Pensions Dashboards implementation is data matching. This capability is at the heart of the Dashboards Programme—defining how pension trustees and providers choose to match Find Requests with their pension records.

From our long experience working with pension records, we know that data matching could be the single biggest point of failure, or hopefully success, for the entire dashboards initiative. Working with their technology and specialist data partners, pension trustees and providers now can, and must, understand the accuracy of key personal data items, right across their membership or customer base, and implement appropriately sophisticated match criteria.

From Pensions Dashboards: Getting to the heart of matching research report

Matching—optimising the critical element of PDP

In order for dashboards to match people to their pensions, there needs to be clear criteria on the data that will be used to do this – particularly to ensure that personal details held by pension schemes and providers are accurate.

Chris Curry, Principal, Pensions Dashboards Programme

Defining matching criteria is a challenge.

Matching pension holders to pension records with a high degree of accuracy—such that almost all pensions are correctly connected to their owners—is the party-trick at the center of the Pensions Dashboards concept.

Because of data quality issues such as data entry errors, incomplete or missing data, or unverified data, every match request has three possible outcomes: Full Match, Possible Match or No Match.

In order to maximise match accuracy and minimise resources consumed by searches, schemes need to define the criteria they will use to determine what match category each Find Request will fall into when checked against each data record.

2 criteria for determining a good match policy: Focus and Coverage

Overall, these sets of matching criteria for Full and Possible Match can be thought of as your match policy. And all good match policies have 2 traits—focus and coverage.

The importance of focus in match criteria

Focus is an important aspect of defining your match policy because criteria that are too broad will create false matches and eat up resources.

The key question to ask when it comes to focus is—are the match criteria you use as a test sufficiently tight to avoid generating too many incorrect responses? For Full Match this means your criteria need to be very well targeted so that you can be certain beyond a reasonable doubt that the pension record in question does indeed belong to the member or policyholder in the Find Request—which is critical because you're about to release pension details to this person.

In the case of Possible Matches, you won't be releasing pension details, so you don’t have to be as concerned about sending a Possible Match response to the wrong person. However, it still makes sense to keep your criteria focused because too many Possible Match responses will tie up admin resources and can confuse pension holders.

The importance of coverage in match criteria

The second criteria that is important for creating an effective matching policy is coverage—meaning, what percentage of your members or policyholders will actually find their pensions based on at least one of your match criteria?

In many cases fuzzy comparisons can be used to increase the likelihood of a match. These types of comparisons recognize when numbers or letters in a field are close, or have been transposed, omitted or duplicated—for instance, a record in which a National Insurance Number (NINO) has 2 digits that are swapped, but the Last Name and First Name are correct could constitute a possible match.

What this means for determining a matching policy is that you can choose whether to have tight criteria such as first name, last name, national insurance number, where a hit on the names and a fuzzy match on the NINO likely means you have the right person. Or you can open it up so that you judge based on first name, last name and date of birth—understanding that you will get more possible matches because there could be several John Smith’s born on 03-09-75.

In some cases, it makes sense for a data controller to have as many as four or five Full Match criteria and a similar number for Possible Match. This increases coverage because the Find Request only has to match one of the criteria. However, you need to be sure that you define each of your criteria such that they don’t create false matches.

Findings from Pension Fusion matching research with 1,000,000 records

The Pension Fusion team carried out research in which we tested the real personal data of a million members across a range of schemes. We used tracing agency data to create simulated Find Requests and attempted to match to the sample data records. Our conclusion was that there is no one-size-fits-all approach, and most schemes need multiple match criteria to achieve the desired level of coverage.

Optimising match criteria so that 99% of members find their pensions

The ultimate question schemes should ask is, just how complete can matching coverage be?

In our research we found that a good set of matching criteria, with an appropriate balance of focus and coverage can increase Full Match rates to over 99% and a Possible Match rate of just over half a percent. And most importantly, the No Match rate was virtually zero. In raw numbers that means that out of 1 million members tested, less than 300 were unable to find their pensions and there were only 6000 Possible Matches.

Prior to optimizing the match criteria there had been 10,000 members without any matches and more 30,000 Possible Matches.

More on the details of those tests, match criteria, testing and data cleaning can be found in the Equisoft and ITM Pensions Dashboards: Getting to the heart of matching report.

As you prepare for PDP staging it is crucially important that you clean your data and define your optimal match criteria for both Full and Possible Matches. Following the guidelines suggested in this article and the further detail contained in our Pensions Dashboards: Getting to the heart of matching report will save time and resources, while ensuring that the maximum number of pension holders are able to find all of their assets.

For more insights into PDP readiness in this Pensions Dashboards Readiness: Essential Steps and Supporting Tools webcast, featuring Equisoft Product Manager David Poynton, ITM Director Maurice Titley, Chris Curry, Principal at PDP and Joanne Eldridge, Pensions Dashboard Industry Engagement Specialist, The Pensions Regulator

Subscribe to our insights newsletter

Related Articles

Gears Co Working 1920x1080


Strategies and Resources for Optimising Pensions Dashboards Programme Matching Criteria

Discover strategies for optimizing data matching in the Pensions Dashboards Programme (PDP).
Read Article


Pensions Dashboards Readiness: Essential Steps and Supporting Tools

During this webinar, our Pension Fusion Product Manager, David Poynton along with guest speakers will walk through the industry toolkits available to Trustees, TPA’s & Providers.
Watch Webcast
Electronic Transfers Review


First Half of 2023 Review of UK Investment Transfer Market

Find out what happened in the UK Investment Transfer market in the first half of 2023
Read Article