This is the latest in an occasional series of blog posts about the campaign for open address data in the UK. Previous posts are listed at the bottom.
In this post I raise questions about the licensing basis on which the Office for National Statistics publishes postcode data linked to unique property identifiers.
Background
In my previous two posts, I wrote about the publication by Leeds City Council of a 2024 list of Council Tax bands of all properties in Leeds. This publication of address data followed FOI and re-use requests that I submitted in October.
Originally, the address data published by Leeds included postcodes, but the Council removed that field following an intervention by Ordnance Survey.
I produced a geocoded version of the Leeds address dataset by adding coordinates from the ONS UPRN Directory (ONSUD). I also appended postcodes from ONSUD, in place of those removed by the Council.
ONSUD is one of two address data products that the Office for National Statistics (ONS) publishes every six weeks via its Open Geography Portal. The other is the National Statistics UPRN Lookup (NSUL). Both data products contain lists of UPRNs (Unique Property Reference Numbers) matched to geographic coordinates, postcodes, and identifiers for a range of current statutory administrative, electoral, health and other statistical geographies.
ONSUD and NSUL include UPRNs for all properties in Great Britain. The data, including postcodes, is published for re-use under the Open Government Licence. ONSUD and NSUL do not contain address fields, but the UPRN and postcode fields are derived from AddressBase®, Ordnance Survey's authoritative geocoded address data product.
AddressBase® includes address data from Royal Mail's Postcode Address File (PAF), as well as from other sources. Neither AddressBase® nor PAF is open data, but both are available for re-use on commercial terms and are free to use (with restrictions) by UK public sector organisations under the Public Sector Geospatial Agreement (PSGA).
Royal Mail and the UK Government have consistently maintained that PAF, and in particular the list of postcodes in PAF and AddressBase®, is the intellectual property of Royal Mail.
Why is ONS allowed to publish postcodes as open data?
As a re-user of ONS's address data products, I was interested to understand why ONS is able to publish postcodes appended to UPRNs as open data, without any apparent objection from Ordnance Survey or Royal Mail, whereas Leeds City Council – and presumably other local authorities – cannot.
I also noticed that, according to ONS's user guides, the postcode lists used in editions of ONSUD and NSUL are as much as 17 months out of date. This surprised me. The postcodes are arguably the "unique selling proposition" of ONSUD and NSUL, since all of the other information they contain can be reproduced and matched to UPRNs using ONS's other open geographic products. Why is ONS not publishing more timely postcode data?
Early last month I sent a Freedom of Information request to ONS, asking for the following information:
- The currently in effect version of any agreement between ONS and Ordnance Survey that relates to the inclusion or use of postcodes in ONS address products, or the production of that postcode data. I don't know what form this agreement might take – it might be a legally binding licensing agreement, or it might be a non-binding memorandum or in correspondence. The agreement might refer more generally to use of AddressBase data – if so, please consider that to be within scope of my information request provided that the AddressBase data includes postcodes.
- Any unpublished guidance or technical documentation that relates to the inclusion or use of postcodes in ONS address products.
- Any other information that explains the currency of the postcode data included in ONS address products. (I note that the latest releases of ONSUD and NSUL contain postcodes as included on the May 2023 version of Code-Point Open, and am interested to understand why the postcode data is so old.)
ONS published its response just before Christmas.
ONS's explanation of the licensing basis for use of postcodes in ONSUD and NSUL
ONS says in its response that there is "no formal agreement with regards to the inclusion or use of postcodes within ONS' UPRN products." ONS gives the following reasons:
- Whilst the UPRNs used are derived from AddressBase® data, ONS publish the UPRNs with no other attributes, in line with the UPRN open identifiers policy.
- The postcodes allocated to the UPRN, whilst derived from AddressBase®, are checked against the OS Code-Point Open (Postcode) product to ensure only open postcodes are published. Any postcode attributed that is not appearing in Code-Point Open is deleted, thus complying with open data licencing.
This differs from a previous ONS statement, in the user guides for the editions of ONSUD and NSUL released in October 2020 through October 2021, that "Postcodes not present in Code-Point Open have been redacted under the terms of our agreement with Ordnance Survey." That statement seemed to imply a particular agreement between ONS and OS.
(We know from earlier FOI responses that ONS signed a Memorandum of Understanding with OS in 2018, which lapsed in 2020 and has not been renewed. But that MoU was at a high level and did not mention individual data products.)
What ONS now seems to be saying is that the inclusion or use of postcodes in ONSUD and NSUL is based on a combination of (a) the rights ONS has as an AddressBase® licensee under the Public Sector Geospatial Agreement (PSGA), (b) ONS's interpretation of Ordnance Survey's Open Identifiers Policy, and (c) public rights to re-use data in OS Code-Point Open under the Open Government Licence (OGL).
OS's Open Identifiers Policy was implemented in July 2020. Among other things, the policy allows PSGA members to use non-open attributes in AddressBase® to match their own property-related data to UPRNs – and then publish that data, with the UPRNs and their geographic coordinates appended, on the terms of the OGL or another open licence.
The postcode is one of a number of AddressBase® attributes that can be used for matching purposes, but the policy does not permit PSGA members to append and publish the postcodes themselves as open data. Postcodes are "Address data" for the purposes of the PSGA licence. As Ordnance Survey said in its recent correspondence with Leeds City Council: "Address data from OS and Royal Mail is only made available under licence conditions, which does not permit open publication, recognising the various copyright and database rights."
According to my reading of the Open Identifiers Policy, it is immaterial that the list of postcodes has been winnowed to include only those that are separately available as open data in OS Code-Point Open. The postcodes used in ONSUD and NSUL are still an AddressBase® attribute because AddressBase® is the only source of the linkage to the UPRNs.
Unfortunately, ONS also says in its FOI response that it does not "have unpublished guidance or technical documentation which relates to the inclusion of the postcodes specifically, only working knowledge by individuals within the team." This suggests ONS may not have formally documented the provenance of third-party data in ONSUD and NSUL or the legal rationale and permissions for its re-use.
Why is the postcode list in ONSUD and NSUL so out of date?
As I noted in my FOI request, the latest releases of ONSUD and NSUL (November 2024) contain postcodes as included on the May 2023 version of OS Code-Point Open. That's 17 months behind the latest available version of Code-Point Open, which is updated quarterly.
ONS says:
We usually thrive [sic] to ensure that the postcode information is as current as possible, using the latest OS AddressBase and Code-Point Open product releases to update our products. Unfortunately, due to a current resourcing shortage, the ONS Geography team has not been able to maintain that currency. We always ensure however that we clearly state which source of data we use to avoid any confusion …
We can confirm that the November 2024 version of Code-Point Open will be used in the first UPRN directories issued in 2025.
I am sceptical of this explanation, for a couple of reasons.
If ONS is extracting an updated list of UPRNs from AddressBase® every six weeks, the additional resource required to extract the corresponding postcodes at the same time seems minimal – even if the ONS Geography team needs to remove new postcodes that have not yet appeared in Code-Point Open.
And if ONS is short of resources to update the postcodes, that shortage seems to be more than "current". According to the historic user guides, ONS has only used two versions of Code-Point Open in the past three years – it has been updating the postcode list on an 18-month cycle that is about to begin again with the next editions of ONSUD and NSUL.
My suspicion is that ONS has been "nerfing" the postcode lists in ONSUD and NSUL, perhaps to allay OS concerns that more current information would make these open products competitive with AddressBase® for some use cases.
Comments
This post isn't intended to dissuade ONS from publishing postcodes in ONSUD and NSUL – indeed, the availability of this information is invaluable, given that an open national dataset of UPRNs matched to postcodes is not available from any other source.
My concern is that the licensing status of the third-party data is not adequately documented, and that this may present difficulties for re-users in future. While ONS and Ordnance Survey are both public bodies, they have quite different organisational cultures and approaches to open data – "open by default" is intrinsic to ONS's business model, whereas OS has strong commercial priorities and has historically fought tooth and nail to limit the amount of open data it is obliged to publish.
In theory, if ONS cannot evidence formal permission to re-use postcodes from AddressBase® in ONSUD and NSUL, Ordnance Survey or Royal Mail could raise legal objections that would affect re-users.
Of course, it could be that my analysis is wrong and that ONS is correct in its apparent belief that no specific agreement is required. But that raises a different question: if ONS can rely on the Open Identifiers Policy as sufficient authority to publish postcodes linked to UPRNs, why is that option not also available to other PSGA members such as Leeds City Council?
I have asked ONS for an internal review of its FOI response, and will update this post if I receive any further clarification.
Previous posts related to the campaign for open address data in the UK
Open address data published by Leeds City Council – Ordnance Survey intervenes to protect Royal Mail IP (24 November 2024)
Open address data published by Leeds City Council shows the way to a national open address file (16 October 2024)
Whatever happened to the UK Government's Open Address Register programme? (1 October 2024)
FOI release: Royal Mail's view on a House of Lords proposal for open address data in the UK (15 August 2024)
Open addresses in the House of Lords – what happened? (28 March 2024)
Thoughts on campaigning for open address data in the UK (30 June 2023)
UK address data: a primer and bibliography (last updated 30 November 2022)