Automatic registration pilots evaluation

We have received some data from canvasses which shows how many automatically registered electors were removed during the canvass. However, at time of publication it has not been possible to include final canvass data. The Welsh Government should assess this data in order to satisfy themselves that our conclusions here are supported.

The data matching process can be time consuming and labour intensive with significant manual work required after the automated matching. In addition, there are costs associated with the volume of invitations to register (ITRs) which are sent out to electors as a result of the data matching. The experience of the pilots will not necessarily be representative of all local authorities in Wales and the Welsh Government should consider what additional resourcing would be needed by EROs across Wales. 

We cannot draw conclusions, from the pilot costs, about the total cost of a Wales-wide implementation of automatic registration. Some costs, such as Electoral Management Software (EMS) development, would not need to be repeated in full. Others, including the resourcing point above, could vary significantly for different local authorities. A final assessment will also depend on the timing and frequency of the registration processes which are put in place.

The Welsh Government should consider the optimal timing and frequency of an automatic registration process – particularly in relation to other aspects of election teams’ work, such as the annual canvass and scheduled elections. This could be carried out each year between the conclusion of the annual canvass and ahead of any scheduled May elections. In years without scheduled polls it would be beneficial also to carry out an exercise ahead of the canvass.

The Welsh Government should consider a reduction in the notification period from 60 days. A shorter period would offer more flexibility around the timing of an automatic registration exercise. It could also mean less opportunity for elector confusion about whether they are registered to vote or not – particularly in the event of by-elections.

EROs should be required to send the first notification of automatic registration in hard copy to the elector’s address, rather than by email. In the event of inaccurate matching, a contact with the address offers the opportunity for existing residents to flag this to the ERO.

An ITR should be included with the notification letter. This would streamline the process, help to mitigate voter confusion and could reduce costs. Auto-registered electors must be sent an ITR at some stage because the ERO cannot be sure, from local data matching, if they should also be registered for UK general elections.

The Welsh Government ensure there is appropriate guidance on data quality and usage as part of a roll out of automatic registration. Guidance could give EROs additional confidence about which data sources they can or should not use beyond examples like council tax.

Successful delivery of any electoral registration process relies on the functionality of EMS systems. Development will be needed for those EMS providers not involved in these pilots and there are also lessons to learn from the pilots. The Welsh Government implementation plans will need to agree a permanent register ‘marker’ for automatically registered electors as well as exploring improvements to the data matching processes – particularly in relation to how results from different datasets are combined.

The Welsh Government should consider how a form of automatic registration introduced in Wales is best implemented in order to allow for efficient future integration with any UK-wide system. The UK Government should engage with and take account of the Welsh Government’s plans as they develop a UK-wide approach. Several automatic registration systems in the UK are unlikely to be value for money, and risk confusion for voters and EROs.

* This is the minimum number matched with two data sources – some additional matches were manually added where the system failed to correctly identify them.

Note: These figures represent the state of play in later October / mid-November. Some data, e.g. the number of people who asked to opt out or the number registering for the parliamentary franchise, could be subject to change. 

We cannot make a precise assessment of the accuracy of the data matching and automatic registration process, e.g. what proportion of the people contacted and registered were resident at that address, at the time. Although the need for two dataset matches clearly increases the likelihood that they are current residents.

The best evidence from these pilots is in the proportion of the automatically registered electors that were later removed during the annual canvass. This is also not a precise measure as it requires someone at a property to respond to the canvass indicating that a listed elector does not live at the address – we know that not everyone will do that. However, it does give a good sense of the overall level of accuracy in the matching.

Positively, by early to mid-November 2025, a small number of electors added during automatic registration had been marked for removal as a result of the annual canvass – 17 in Powys and 66 in Gwynedd.

Given the timing of this report, it has not been possible to include final data on removals from the annual canvass. The Welsh Government should request this data from the pilot areas in order to satisfy themselves that our conclusions here are supported.

People identified using education data or more than one of the other datasets, were sent a notice of registration. This informed them that they would be added to the register in 60 days and be eligible to vote in Senedd and local government elections. Those who wished to opt out were asked to contact the pilot authority within 60 days.

The communications used in the pilots were designed and tested by the Behavioural Insights Team with the aim of making them as clear and understandable as possible for electors.

The pilot legislation allowed for these communications to be sent either by post or by email. However, in the pilot all of the notifications were sent by post. As part of any wider implementation, we recommend that EROs are required to send the first notification in hard copy to the elector’s address, rather than by email. In the event of inaccurate matching, a contact with the address offers the opportunity for existing residents to flag this to the ERO.

Electors automatically registered through this pilot were added to the register for devolved elections in Wales (Senedd and local government). They could not be automatically registered for reserved elections (UK general elections or Police and Crime Commissioners (PCCs)).

Feedback from local authorities was mixed on the practicalities of managing the different franchises if automatic registration was rolled out. Some felt that adding electors for only one franchise, when they may be eligible for both, was an unhelpful complexity. Others felt that the complexity, in terms of different franchises, already exists. However, they did think it would be important for their EMS systems to be able efficiently to manage these requirements.

This means EROs being able to clearly identify the automatically registered electors on their registers. An electoral register for an area is a single database with ‘markers’ placed against individual electors to allow different eligible electorates to be generated, depending on the poll in question.

Some electors, e.g. some EU citizens, are already flagged (with a ‘B’ or ‘G’ marker) to show they can vote in Senedd, local and PCC elections but not in UK general elections. However, ahead of these pilots, there was no existing marker which could be used to accurately distinguish those who were automatically registered. The need to agree a suitable marker caused some delays in the pilots. If automatic registration is rolled out, a permanent approach should be settled. This could be the continued use of the temporary ‘J’ marker used in the pilots.

Importantly, and unlike the EU citizen markers, this marker will tell EROs that they may be missing information about the elector – specifically their nationality and that they may still be eligible to register for other polls (and should be contacted).

Ultimately, avoiding confusion for voters is crucial. Without the effective management of the different franchises there is a risk of voters not being made aware of the need to register for upcoming election. By-elections present a specific challenge because they can take place at almost any time in the year. This would be exacerbated if the by-election was for a reserved poll where the automatically registered voters would definitely not be registered for that election (but may believe they were).

The notifications sent in the pilots clearly told voters that they would need to register separately for certain other elections but did not include an ITR. In some areas, reminders were sent during the 60 day period which also included information for electors on the possibility of registering for the parliamentary franchise. But no ITRs were sent out to these electors within the 60 day period.

The auto-registered electors must be sent an ITR at some stage because the ERO cannot be sure if they should also be registered for UK general elections. In any future roll-out of automatic registration, combining the notification and ITR would streamline this process, help to mitigate voter confusion and could reduce costs.

The reaction of the people contacted about automatic registration was an important aspect of the pilot. We know, from our previous research, that over half of people in Great Britain think automatic registration would be an improvement on the current registration system. Positively, the local authorities received limited inquiries and few complaints. As above, we know that not everyone contacted will have fully engaged with the information provided and that may be one factor in the lack of response. However, the limited response does indicate the absence of a widespread concern about the approach of registering electors without an application.

Across the three pilots, between 0.2% (Gwynedd) and 4% (Newport) of those sent automatic registration notifications asked not to be added to the registers. There were some common elements within this group:

• 
  For the pilots using education data, and therefore registering attainers aged 14 – 15, they did receive some contacts from concerned parents/guardians. In many cases, this was parents checking whether it was correct that a 14- or 15-year-old should be added to the electoral register. Often, following an explanation from the electoral services team, the query was resolved and the elector was registered. In other cases, they opted out of the automatic registration process.
• The pilots also received a small number of queries from people who did not want to appear on the electoral register for safety reasons. They were understandably concerned by the information that they were about to be registered. The electoral services teams were able to offer the option of using the existing anonymous registration process, rather than being automatically added. However, in nearly all cases, the individuals were not reassured by the anonymous registration provisions and preferred to remain unregistered.
• Some people did raise more general concerns. They were often strongly held, e.g. where they actively wished to remain unregistered or did not like the idea of being identified through other data sources. But this was only proportion of the small number overall who asked to opt out.

While overall volumes of contacts were relatively low, and the local authorities did not increase their staffing as part of the pilot, handling these enquiries did require staff time. As part of any roll out, individual authorities would need to assess their ability to handle enquiries depending on the volume of automatic registration letters they were sending out and their existing staffing situation.
 

Currently, EROs must send individual ‘invitation to register’ forms (ITRs) to potential electors identified by, for example, direct contact from individuals or by local data matching. A single match against a dataset was not sufficient to automatically register an individual in these pilots but it is enough to require an ITR to be sent.

Legally, once an ITR is sent EROs must also take additional steps to encourage a response, including sending postal reminders, making a visit to the property and/or sending a personal email follow up.

Where pilot areas issued ITRs to these electors during the pilot period, they confirmed that this was a resource intensive activity both in terms of issuing the ITRs and in handling responses. For example, Powys issued nearly 7,000 ITRs to potential electors in the pilot. This would have been over 12,000 ITRs but they sent an initial letter to this group which encouraged many to register online and also identified inaccuracies in the data matching.

This therefore has the potential to be a significant commitment of resources for all EROs if automatic registration is rolled out in Wales.

The Welsh Government will need to consider how any automatic registration process will interact with these existing commitments in relation to ITRs and ensure that EROs have the resources needed to adapt to this increased workload.

Any implementation of automatic registration should consider the timing and frequency of data matching and elector contact, particularly in relation to other aspects of election teams’ work, such as the annual canvass and scheduled elections. It is also important that it is clear to EROs whether this process is to be mandatory or at their discretion.

The pilots tested a single process of matching and registration within a year but depending on the availability of data and the efficiency of the process, it could happen more frequently. A single process could be carried out each year between the conclusion of the annual canvass and ahead of any scheduled May elections. In years without scheduled polls it would be beneficial also to carry out an exercise ahead of the canvass.

However, this would only be viable if the matching processes were as efficient as possible and if EROs have sufficient resources to conclude the work within the available time:

• Automatic registration activity needs to be completed, and not overlap with, the annual canvass: experience from the pilots indicates that running the automatic registration process into the annual canvass period creates issues. For example, whether and how to represent automatically registered electors on the household canvass forms. There are advantages and disadvantages to how the canvass could be used. A canvass following automatic registration would offer another opportunity to tell automatically registered electors that they are registered for certain elections and encourage them to register ‘fully’. A process following the canvass, and ahead of scheduled polls, could pick up new electors that were missed by the canvass or who have moved more recently.
• The process must also be completed well ahead of scheduled polls: In order to avoid elector confusion about whether they are registered or not, the process (including the period between notification and registration) should be completed well in advance of a scheduled election. Clear communications and an agreed approach to handling electors in relation to any unscheduled by-elections are also needed.

The question of timing also depends on the approach to the notification or ‘waiting period. In these pilots the legislation required there to be 60 days between the issuing of a notice of registration and the elector being added to the registers.
We recommend that the Welsh Government should consider a reduction in the notification period. A shorter period would offer more flexibility in relation to the timing of an automatic registration exercise. It could also mean less opportunity for elector confusion about whether they are registered to vote or not – particularly in the event of by-elections.

While the 60 days did not present significant issues for the pilot areas, there is also no specific reason why this period should be 60 days – it does not relate to existing electoral registration processes, e.g. registration reviews or the rolling registration periods.

Reducing the period could increase the possibility of adding electors to the registers incorrectly (where a response could be received to say the elector is not resident) or when they do not want to be (where an opt-out response could be received). However, the pilots noted that the majority of responses tended to be received in the first 2 – 3 weeks of the 60 days.

Alongside the wider decision about the scheduling of automatic registration processes, the Welsh Government should consider a reduction in the notification period. This could bring it into line with existing review periods (14, 28 or 42 days depending on the specific circumstances) and also allow EROs to align the end of the period with a rolling register update.

23% (15 people) of those who initially signed up also responded to the follow up contact but refused to provide further personal details. This was despite the significant work of the civil society organisations to recruit them initially. Also, the majority of those who did respond, and provide sufficient information for data matching, needed multiple messages and calls to reach that point. This underlines the challenge presented by hard-to-engage groups – successful engagement often requires labour intensive, sustained and multi-step work. This may also have been made more difficult in the pilot by the lack of an upcoming electoral event to focus the engagement around.

58% (25 out of 43) of those who provided details could not be matched with sufficient data to be automatically registered – while this was a disappointing result in terms of registration numbers, it does address one of the aims of the pilot in showing that the outreach work was identifying people who would not otherwise be picked up by locally held data.

Many people were not aware of whether they were registered or not. 42% of the contacts who also filled in a Small Axe feedback survey were already registered to vote but over three quarters of them believed that this process would have been their first time registering to vote.

The Welsh Government should consider the optimal timing and frequency of an automatic registration process – particularly in relation to other aspects of election teams’ work, such as the annual canvass and scheduled elections. This could be carried out each year between the conclusion of the annual canvass and ahead of any scheduled May elections. In years without scheduled polls it would be beneficial also to carry out an exercise ahead of the canvass.

The Welsh Government should consider a reduction in the notification period from 60 days. A shorter period would offer more flexibility around the timing of an automatic registration exercise. It could also mean less opportunity for elector confusion about whether they are registered to vote or not – particularly in the event of by-elections.   

EROs should be required to send the first notification of automatic registration in hard copy to the elector’s address, rather than by email. In the event of inaccurate matching, a contact with the address offers the opportunity for existing residents to flag this to the ERO.

An invitation to register (ITR) should be included with the notification letter. This would streamline the process, help to mitigate voter confusion and could reduce costs. Auto-registered electors must be sent an ITR at some stage because the ERO cannot be sure, from local data matching, if they should also be registered for UK general elections.

The Welsh Government ensure there is appropriate guidance on data quality and usage as part of a roll out of automatic registration. Guidance could give EROs additional confidence about which data sources they can or should not use beyond examples like council tax..

Successful delivery of any electoral registration process relies on the functionality of EMS systems. Development will be needed for those EMS providers not involved in these pilots and there are also lessons to learn from the pilots. The Welsh Government implementation plans will need to agree a permanent register ‘marker’ for automatically registered electors as well as exploring improvements to the data matching processes – particularly in relation to how results from different datasets are combined.

The Welsh Government should consider how a form of automatic registration introduced in Wales is best implemented in order to allow for efficient future integration with any UK-wide system. The UK Government should engage with and take account of the Welsh Government’s plans as they develop a UK-wide approach.