Skip to main content
Success Story | Pensions Briefing

The benefits of data digitisation

A case study of the Philips Pension Fund


By Adrian Holmes | July 30, 2015

Adrian Holmes, Pension Manager at Philips, shares his experience of improving data quality using a novel approach that saved Philips an estimated £8m.

How the issue arose

The Philips Pension Fund is a large, long-standing defined benefit plan. It has a level of historical complexity that is typical for many such plans, with several distinct sections and non-standard benefits for some members transferred in following business acquisitions as Philips’ UK business evolved over time. The Trustee had carried out a programme to improve and maintain electronic data quality spanning several years. However, when the opportunity arose to secure benefits with an insurer (a ‘buy-in’), the Trustee realised it needed to review the electronic data in more detail so that it could benefit from the most attractive market pricing.

While preparing data for the Fund’s first buy-in exercise, we identified a member with a special pension guarantee recorded in the supporting files but not on their electronic record. The insurer therefore asked the Trustee to ensure the electronic data fully reflected any members entitled to this pension guarantee. This represented a challenge, as we would have needed to review all of the supporting files for the entire pensioner membership – some 15,000 members!

We also identified that this was one of several historical details not held on electronic records. This is a common situation for many schemes and may not be a concern for ongoing administration, as long as there is a process in place to pick up the detail, for example when benefits come into payment. However, the insurer required all data to be provided in an accessible electronic form. If this had not been possible, the insurer would have increased its cost quoted for the buy-in to reflect this data uncertainty.

The ‘traditional’ manual approach

Traditionally, the only solution would have been a manual file review of the supporting scanned images, which we estimated would have taken at least 1,000 man-hours and up to 1,500 hours. Hence we would have either had to delay the buy-in for several months, or pay a higher insurance premium.

Faced with this choice, we were keen to investigate a quicker and more cost-effective solution. I discussed the issue with the Towers Watson specialist data team leading our data improvement programme.

The solution: data digitisation

We worked with Towers Watson to develop a solution which involved making the data fields on the scanned images digitally readable so that they would become searchable in bulk (see Figure 01). Data digitisation techniques have rarely been used in the past, but recent improvements in technology now provide trustees and sponsoring employers with new options.

There were several areas where we thought this solution would be beneficial. Initially, as a trial of this approach, we focused on the pension guarantee that had highlighted the problem in the first place. We were able to analyse some 1.5 million documents swiftly and accurately. Where the document search indicated the possibility of a special guarantee, a focused file review was completed by the team and we identified a further 10 members with the special guarantee. Once the detailed set-up work had been completed, the actual data extraction took only 20 hours, compared with at least 1,000 hours for a manual review, as illustrated by Figure 01.

Figure 01: Comparison of time taken to extract the required data


The insurer was satisfied that the process used was robust and removed uncertainty about the guarantee from the data. Any upward adjustment to the insurance premium was avoided.

The data digitisation process

1. Assess and identify – Review the original records (such as paper files or scanned images); identify target data fields.

2. Configure and extract – Scan (for paper records) and process the documents. Configure the software to identify data items of interest. Extract the target data into usable formats.

3. Review and load – Manually review any ambiguous or unreadable data. Verify results by applying risk-based checks and controls.

The end result is an ordered set of data in an electronic, searchable format, and extracted data which can then be loaded directly into the administration system.

Follow-up work

Having successfully addressed the pension guarantee, we then used data digitisation to improve the data available in electronic form. Two examples include:

  • Funeral benefit – Some of our members are entitled to a funeral benefit, which is documented on a form held as a scanned image. Using data digitisation, we were able to construct an electronic list of the relevant members. Without data, the insurer had originally priced this benefit assuming that all members were entitled to it.
  • Contingent spouses' pensions – Spouses' pensions had always been calculated when they came into payment. Using a digitised approach, we were able to extract the data required to calculate contingent spouses' pensions in advance for 3,000 pensioners.

The total saving in the buy-in premium from these exercises has been estimated at £8m.

Benefits of this work to the Trustee

  • Significant savings in the buy-in premium - As described above. Providing more detailed data gave the insurer more certainty. These savings covered the cost of these exercises many times over.
  • Quicker results – Although time and effort was initially required to set up the process, this was more than outweighed by the huge reduction in data processing time compared with a manual approach.
  • More efficient ongoing administration - The improved data quality will enable the administrator to respond more quickly to member enquiries.
  • Reduced risk – As a trustee board, we had confidence that a robust, automated process had been followed, reducing the risk of human error involved in a manual process. Computers don’t get bored, even with millions of images to work through and we had reassurance in the results produced.

Providing more detailed data gave the insurer more certainty. These savings covered the cost of these exercises many times over.

Could this work for others?

We are pleased that what began as a project to address a specific issue led to further financial savings in related areas, not only for the buy-in but for the ongoing administration of the Fund. For any pension scheme that holds data in paper formats or scanned images, I would recommend considering whether this approach could yield similar benefits.

For any pension scheme that holds data in paper formats or scanned images, I would recommend considering whether this approach could yield similar benefits.

To discuss how data digitisation could benefit your scheme, please contact your Willis Towers Watson consultant or Phil Titchener.

Adrian Holmes is the Pension Manager at Philips. In this article, he outlines the novel approach that the Trustee of the Philips Pension Fund took to improve its electronic data and the impact this had on the Fund.

Title File Type File Size
Trustee Briefing July August 2015 PDF 2.7 MB
Related content tags, list of links Article Pensions Briefing United Kingdom
Contact Us