Thoughts on campaigning for open address data in the UK

Following is my current thinking on how to campaign for open address data in the UK.

This is a complex matter, but I'll try to be pithy. Readers new to the subject may wish to look at my UK address data primer and bibliography.

Unlocking national address data is the great unfulfilled dream of the UK's open data community. Some might even call it our white whale – a goal that we have chased obsessively but are unlikely to obtain.

Recent interventions include a blog post from journalist James O'Malley, making his case for opening up Royal Mail's Postcode Address File (PAF), and the Centre for Public Data's call for Ofcom to review UK address data.

CFPD founder Anna Powell-Smith and others have also set up a campaign site: Open Address File UK.

What is the 'ask'?

The core objective should be open release of an authoritative UK-wide address dataset that includes full postal addresses, UPRNs, and point coordinates, accessible online as a bulk download and regularly updated.

But, given that most of the commercial value in the national dataset is in the addresses and coordinates, there is no reason to limit ourselves to that simplified set of fields.

We should also expect the addresses to be attributed with metadata, such as the local authority reference and most recent update date. The address list should also include properties that are not postal delivery points, but have UPRNs, such as churches and park pavilions.

Open address data should be suitable for all readily identifiable use and re-use cases. That must include the existing purposes for which the data is maintained, and the needs of existing public sector users and commercial licensees, in addition to the many legal purposes and business models that are unachievable under current licensing conditions.

We should aim to unlock national address data in its entirety, rather than a subset or simplified data product. Address data will only function effectively as digital infrastructure if the full dataset is accessible across markets and sectors, and at all levels within supply chains, as well as to individual members of the public.

In practice, that means we should campaign for open release of both the Royal Mail's Postcode Address File (PAF) and Ordnance Survey's AddressBase products. PAF is more timely, but only contains postal delivery addresses and does not include either coordinates or UPRNs.

Following from that, the ideal would be read-only public access to the database that underpins the National Address Gazetteer, so that re-users could pull the data they want directly from the central source.

Don't try to re-engineer the address data system (yet)

I am very dubious about the idea that UK address data can or should be maintained on a collaborative model that resembles OpenStreetMap or Wikipedia, or that there is any pressing need to reorganise the current governance of the address data system.

Our focus should be on making the existing address data products directly available for additional re-use.

The address maintainence system itself is mature and has a complex ecology. Some of the terminology has changed, but this diagram from the 2014 Neffendorf Report remains broadly accurate:

It may be that open release of address data will produce savings within that system – for example, by removing the costs of administering licensing agreements and of keeping the data secure from unauthorised use. It may also make it easier for users to detect and report data quality issues.

But I think that if anybody in the wider data community has ideas for reforming the systems or governance for address data, the onus is on them to demonstrate that expertise – as an activity parallel to opening the data, rather than a distraction from that objective.

What does the business case look like?

We should expect greater transparency from Government about the current costs and commercial arrangements underpinning the management of address data. All of the parties involved – the Geospatial Commission, Ordnance Survey, GeoPlace, the Local Government Association, and Royal Mail – are secretive about the revenue flows and the extent to which the costs of maintaining quality address data are offset by income from licensing for re-use.

Address data is essentially a state-run monopoly in the UK, with no viable outside competition – so there is no real justification for commercial confidentiality.

Only when we have that financial information, and know what the business case is for the current approach, should we entertain any challenge to produce a business case for the counterfactual case, open addresses.

(But in the meantime we can look to international examples. See Peter Wells's blog post on countries that have already opened their address data.)

I have a few working assumptions. Current address data arrangements, including the licensed data products, are designed primarily to support the postal delivery system and fulfil public sector purposes such as local planning and Council Tax assessment. The associated revenue flows are mainly a distribution of government funding. Even Royal Mail's role in the address system is driven mainly by the Universal Service Obligation, which is in the gift of Government.

Public sector bodies have free access to national address data (for a limited range of purposes) under the Public Sector Geospatial Agreement and the Royal Mail Public Sector Licence.

In relative terms, the contribution of revenue from commercial licensing of address data is small. Low millions, annually? The bold approach would be to legislate for open addresses: make open licensing of PAF a condition of the USO, and eliminate any ring-fenced funding for re-use of AddressBase. But in practice, Government may want to make up the shortfall in commercial licensing revenue, from public funds.

That is only bookkeeping. Government's loss of commercial revenue from address data would be equal to savings by existing licensees, so represent an economic stimulus in the short term. (In the longer term, those licensees would likely face greater competition because the input cost of licensing address data will no longer be a barrier to entry for smaller companies.)

Even if we ignore the potential economic benefits of unlocking address data, the loss of direct revenue is unlikely to have a noticeable effect on public accounts. For perspective, we can consider the annual income that Land Registry receives from fees to register changes to the title of land and property: £307 million in 2021/22. If necessary, loss of revenue from address data could be offset by a small increase in Land Registry fees.

Talking points: broad areas of benefit from unlocking address data

As a thought exercise, consider the economic impact if an organisation published a free API that enabled anyone to search and extract UK address data, and re-use it for any legal purpose.

What would the take-up be, do you think?

It seems self-evident that a multitude of business uses, and savings, flow from that idea. Some of those uses will increase revenues and profits, which translates into additional tax revenue for the Government. (Accepting, of course, that this effect can be difficult to quantify. Data is never the only factor in running a profitable business.)

Currently, such an API is impossible to implement. Even if an organisation was willing to absorb the licensing costs, address data from Royal Mail and Ordnance Survey is simply not available on terms that allow a licensee to give end users unrestricted rights to re-use it.

Lack of availability of open addresses compromises re-use of other public data infrastructure. For example, Land Registry's Price Paid Data and DLUHC's Energy Performance of Buildings Data. Both datasets would be fully open if Royal Mail restrictions on re-use of the address data fields were removed.

On the other hand, Companies House's Free Company Data Product, which contains addresses for all companies registered in the UK, is fully open data – because Companies House simply records the individual addresses provided by registrants, and does not validate them against PAF or AddressBase.

This trade-off between quality and openness undermines the utility of address data across the UK economy, as well as in the public sector and third sector. Any organisation that has collected its own list of addresses has a choice: either accept that the list might contain incorrect or non-standard addresses, or cleanse the list using PAF or AddressBase.

Good practice is to cleanse the list, in the interests of better data quality and interoperability with other datasets – but that means the organisation has to give away some ownership of its address data, and will be restricted in how it can share the cleansed data.

Open availability of an authoritative national address dataset would inevitably lead to higher quality address data throughout the UK, because it would reduce the cost of validating addresses at the point of collection and remove the disincentive to standardise addresses already held.

More talking points: the status quo isn't working

In 2010, the UK open data movement had an early victory in securing the release of a package of basic Ordnance Survey products that included Code-Point Open, a list of postcode coordinates for Great Britain.

Government support for open data has flagged since then, but calls for open addresses have continued. Ordnance Survey, and its current political wing the Geospatial Commission, have offered two concessions to that pressure.

The first is inclusion of address data in the Public Sector Mapping Agreement, now called the Public Sector Geospatial Agreement, which makes address data free at the point of use for public bodies.

We should challenge the Government to demonstrate the success of the PSGA model – with quantitative survey data, as opposed to cherry-picked use cases:

• How many public bodies actually use licensed address data, for purposes other than local planning (where maintenance of address data is a native function)?
• How often do public bodies find that licensing restrictions or security requirements prevent them from making full use of address data?
• Most importantly, how easy is it for individual staff to access address data within their organisations, even when the organisation is signed up as a PSGA member?

Then we can point out how much easier open availability of address data would be, even for PSGA members.

The second OS concession was the 2020 release of UPRNs with their coordinates, as open data. UPRNs are unique identifiers for addresses (and other objects).

OS promoted this release on the basis that public authorities would now be able to publish open datasets that included UPRNs, so that addresses could be matched across datasets even if they didn't contain the addresses themselves. The Government announced that its Open Standards Board had mandated the use of UPRNs across government and the NHS.

But how successful has the open release of UPRNs been?

Other than OS's original example of putting UPRNs into EPC data, and the inclusion of UPRNs in local authority datasets unlocked very recently by Scotland's Improvement Service, is there any systemic evidence that publishing UPRNs as open data has improved the re-use of address data?

Few public bodies seem keen to adopt UPRNs in their infrastructure datasets. For example, Land Registry continues to sell a separate Title Number and UPRN Look Up dataset, rather than adding UPRNs to its own register data, and the Valuation Office Agency has not added UPRNs to its Council Tax dataset.

Those of us who analyse geospatial data regularly can find some technical uses for open UPRNs, principally in the form of the administrative geography lookups produced by ONS (under a separate arrangement with OS).

But it is very difficult to make the average business client understand why they can download and make free use of identifiers that represent UK addresses, but have to pay exorbitant licensing fees to make limited use of the addresses themselves. On the face of it, the situation is absurd.

We should also be concerned about the effect of open UPRNs on the quality of address data. If a third-party dataset contains UPRNs, but no addresses, re-users who do not have access to AddressBase have to take it on faith that the UPRNs have been allocated correctly. If there are errors (as there will be in any large dataset, such as the EPC data), those errors can be propagated to other datasets, with no visual sense-check of the address data.

Who can open address data?

The main blocker to open address data is a lack of political will, rather than the economics.

Unlike most public data assets, address data has a mature intermediary market. In addition to Royal Mail and the various public bodies that share in revenue from commercial licensing of address data, there is an entrenched group of OS partners who function as "value added resellers" and benefit from the artificial scarcity of address data.

Culturally, those partners mainly sit within the property sector, a Tory heartland – which means they have a stronger voice in government than most businesses.

The Geospatial Commission is a lost cause – it largely ignored the numerous representations about address data in its 2018 call for evidence on geospatial strategy, and has dedicated itself to shoring up the position of Ordnance Survey and its other data-rich partners.

Similarly, unless Royal Mail collapses or loses its Universal Service Obligation, I doubt Ofcom will find any reason to change the stance on licensing of PAF that it took following its 2013 consultation.

Assuming there is a general election in 2024, we cannot necessarily expect a more receptive attitude from Labour. However, a change of Government will at least provide an inflection point for reconsideration of open addresses and open data policy more generally.

We may get only the usual noises about transparency and innovation. However, we could lay the ground by raising the following points. The new Government should:

• Establish whether the Geospatial Commission has adequately considered the case for open addresses, and demonstrated a stronger case for maintaining the current arrangements. This could perhaps be part of a broader review of the GC's remit and value.
• Require Ordnance Survey and other parties to publish contract arrangements and financials that underpin the maintenance of AddressBase and commercial licensing of address data.
• Ask the Competition and Markets Authority to review the UK's address data market. In 2011, the Office for Fair Trading concluded that the new arrangements (i.e. those now in place) were a monopoly, but that the market was small enough that it did not need to make a referral to the Competition Commission. That market is now larger. There is a strong argument that the re-use of address data would be more extensive in the economy, if not for the current monopoly control. It is also plausible that, were Royal Mail and the LGA not in that monopoly with Ordnance Survey, one or both would be marketing geocoded address data products in competition with AddressBase.

Ultimately, the campaign for open addresses will only succeed if it is picked up by an interested minister with sufficient influence. In the meantime, there are a small number of parliamentarians who have previously taken an interest in the availability of address data. Those that come immediately to mind are Chi Onwurah MP, who is currently Shadow Minister for Business, Energy and Industrial Strategy, and Liberal Democrat peer Richard Allan.

How much of a problem is Royal Mail?

On a final note, I have the impression that some in the open data community think the 2013 privatisation of Royal Mail put the unlocking of national address data beyond reach, because ownership of the Postcode Address File is now outside the public sector.

I think that is overstated as a barrier to open addresses.

Government seems to have given the impression to the Open Data User Group that selling off PAF was an essential ingredient in the privatisation of Royal Mail. A committee of MPs were strongly critical of that decision.

However, a BIS spokesman reportedly said PAF was included in the sale of Royal Mail "because it is an integral part of its operations, not to boost the price".

The prospectus for the sale of Royal Mail only described the PAF as an intellectual property asset – it did not contain any valuation or estimate of future revenues. A subsequent National Audit Office report on the privatisation did not mention PAF at all.

As I've suggested earlier, Royal Mail's maintenance of PAF is driven by requirements to deliver the Universal Service Obligation. Were the USO to pass to another company, the Government would undoubtably expect responsibility for PAF to be passed over also.

Royal Mail is entirely dependent on local government to supply the address data required to keep PAF up to date, under arrangements negotiated between Government and the LGA. The primary source of address data is the local gazetteers maintained by planning authorities – Royal Mail only adds postcodes, and assures the quality of the PAF data products.

It is unlikely that Royal Mail would present an intractable barrier to open licensing of PAF and other UK address data, if the Government was minded to go ahead. Open licensing does not require any transfer of ownership of the PAF data. Royal Mail would no doubt expect a commercial settlement, but Government has most of the leverage.