Computer cleaning data. Photo PashaIgnatov / Getty Images

Checking member data with commercial data cleaning services: NASUWT case study

Accurate member data is essential for a trade union of any size. If the purpose of a union is to build power and improve members’ experiences in workplaces, it is fundamental that it is clear who members are, where they work and how to get hold of them.

This is especially true in the context of statutory balloting – if we don’t know where our members are living, we can’t send them ballot papers; or worse, we send them to the wrong address, which is then counted as a nil return. If other contact details such as e-mail address or phone number are out of date, it can be almost impossible to reach these members.

Unions have been trying a variety of methods to check member data, such as working with activists to data check using peer-to-peer SMS, using surveys, or giving reps data checking tools to use in the workplace. However, making this process more targeted and efficient is always going to be a priority.

Commercial data services

During Spring 2023, there were two significant balloting campaigns running in NASUWT: one in Wales, and one in England. Both campaigns involved a statutory postal ballot and both had the potential to make an enormous impact on members in the workplace. The issues at the centre of both were certainly widely felt and deeply felt, but in order for them to be winnable, the data was going to have to be accurate.

Working with the TUC Digital Lab, we decided to conduct a pilot of a commercial data cleaning service as part of the local campaign that was running in Wales.

These services take a data file and check it against a variety of other data sources. For example, they:

  • check phone numbers to see if the number is connected to a live account;
  • check postal addresses are correctly formatted with the correct postcode;
  • check email addresses are correctly formatted and on live domains;
  • check whether a person at that postal address has moved and left a forwarding address, or notified other agencies that they have moved.

Finding a supplier

Service features

We wanted to find a supplier who could handle all of our requirements – getting checks on as many types of contact data as possible, but crucially the Post Office redirections database.

Many suppliers also offered to check against marketing preference services, but due to the nature of the data we hold, this was not necessary.

Data protection

Keeping member data safe is vital, and there is a severe risk if data is compromised, so it was essential we could be confident that the chosen service could give us security.

In particular we were looking for:

  • a clear statement that the union remained data controller;
  • a clear acknowledgement that no part of the data would be retained after being returned;
  • a statement that the data would be kept within the European Economic Area (EEA);
  • an outline of the process for transferring data to the service provider and back in a secure way;
  • supplier accreditation with info security standards like Cyber Essentials.


Finally, we sought costs on our best-guess scenario for the volume and scope of data we needed to have checked. This was more difficult to judge as some companies charged on a per check basis; others on data returned (meaning the cost would be higher where there were more inaccuracies).

The company we chose (DataHQ) billed on a mix of the two: a flat rate for mobile, landline, email and postal address format checks; and a rate per error found for expired postal addresses.

The process

The actual data cleaning was very straightforward. After we signed the data protection agreement with the supplier, we exported records we wanted to check from our member database into Excel, and uploaded this into the supplier’s secure website.

The supplier checked the data and gave a schedule and quote for processing it. When this was agreed and paid, the supplier made the Excel sheet available for download, with new columns added alongside each of the data fields. The additional column showed the status of the check, and for those addresses where a Royal Mail redirect was in force, the new address was provided.


The data we scanned in the pilot returned errors for postal addresses (both suspected and confirmed moves), mobile numbers, landlines and e-mail addresses. In total, 408 errors were identified in a data set of 413 members. While that sounds staggering, it is less concerning when considering that those records that contained errors usually contained multiple errors, rather than almost all records being out of date.

Running the checks against all channels together cost us the equivalent of around 16p per member record.

The majority of this cost was for the postal redirects and ‘moved away’ checks. As this part of the cost was based on the number of positive results, it may be that repeating the exercise in the future would bring a lower overall cost.

Making use of the data

The pilot involved 413 members. The results from the data cleanse were used in conjunction with self-reported data about ballot returns from members to identify target areas.

What we created was a data-rich picture of what we needed to do to keep the campaign moving. We knew for each workplace a percentage of the membership that had told us they had voted, but now we also had clear information on why they may not have done so. For example, if members were shown to have potentially moved away, they may not have received ballot papers, or if members in a workplace had no valid way to contact them digitally, they may have missed vital information on the campaign.

We were able to use this data to effectively target members and allocate resources. For those workplaces in which lots of members had no valid digital contact details, we knew a workplace meeting was important. For those for whom we had poor address data, we knew we needed to contact members in another way, digitally or via reps and local activists, to get this up to date.

As a part of this, we conducted a follow up with all members for whom we now knew we likely had an out-of-date address, via peer-to-peer texting, which yielded an update rate equivalent to 1% of the balloted population.

All of this allowed us to work in a targeted and effective way and a turnout of 59.4% was achieved, with 92% voting in favour of strike action and 96% in favour of action short of strike, securing a mandate for action and eventually leading to a victory after several days of action.

Where next?

Following the success of this in the pilot, it was decided that a similar methodology would be employed for the national ballot in England.

As we were running a disaggregated ballot, it gave us a warning of workplaces that had high levels of bad address data, and where reaching ballot thresholds would be particularly challenging as a result. This was helpful in allocating resources to the campaign in different schools and had a significant impact on the overall result.

Overall, we achieved a 51.9% turnout across our balloted workplaces, with votes of 88.5% for strike action and 94.3% for action short of strike action. This meant the majority of our balloted members achieved the thresholds necessary for strike action, and was a major factor in winning an improved pay offer from the Government.

Going forward, we are looking to continue to use this style of data cleansing both as part of specific campaigns as detailed above, but also as a part of regular data maintenance. This will allow us to be both proactive and reactive in the way we manage data.

Additionally, in the future, it may be beneficial to consider the benefits of a service with an API that could link directly to a member database to increase efficiency.

Elinor Cheason is an organiser with NASUWT