Case Nos: IL-2023-000080 and IL-2024-000038 - [2025] EWHC 2561 (Ch)

Since the start of these proceedings the defendants have been less than forthcoming about the creation and maintenance of their GetAddress Database. Two orders of Master Pester were required in the claimants’ attempt to secure better information. The orders were not complied with in significant part. Of 61 code files which would have served as a record of what was incorporated into GetAddress only 9 were disclosed. Mr Smith’s evidence in cross-examination was that they were variously irrelevant or that they had been renamed and moved to a different directory which made them hard to find. I doubt that Mr Smith’s answers were both truthful and complete.

RMG and IDDQD provided a long and detailed account of how by correspondence and interim applications they attempted to obtain information from Mr Smith about the source of data which went into the GetAddress Database. The sequence begins with a letter dated 24 November 2014 from Alasdair MacHardy of RMG to Mr Smith stating that the PAF was RMG’s intellectual property, asking for confirmation that the PAF had been used by the defendants, asking from whom the PAF data had been obtained and to whom it had been supplied. Mr Smith replied in an email dated 25 November 2014 with an initial and simple response: no data has been taken from the PAF. He identified publicly available websites as his sources.

I see no point in charting each twist and turn (there were many) in the history of what was done to try to obtain information from the defendants about where the GetAddress data came from and the evolving answers given by Mr Smith in response. IDDQD submitted that Mr Smith had a ‘contemptuous’ attitude to court orders that did not suit him. In cross-examination he described them as ‘annoying’. I am clear that Mr Smith did not take seriously his duty to the court to provide evidence in a complete and candid manner and that this was intended to, and did, hamper the court’s ability to reach an accurate conclusion about the sources of the GetAddress Database.

The defendants’ almost final position before the trial with regard to the initial creation of their database was contained in a revised Response dated 27 January 2025 to a Part 18 Request. The Response was signed by Mr Smith and was stated to represent Mr Smith’s best recollection of how GetAddress was created. He said that 13 sources were used: (i) The Ordnance Survey Code-Point Open dataset, (ii) HM Land Registry Price Paid Data (PPD) (a record of property sales in England and Wales), (iii) Companies House dataset, (iv) the ICO (Information Commissioner’s Office) Register, (v) the Food Standards Agency’s Food Hygiene Ratings Scheme dataset (restaurants), (vi) addresses of GP practices in England, Scotland and Wales, (vii) addresses of dental surgeries in England, (viii) addresses of social housing in England, (ix) an NHS postcode directory, (x) council tax data, (xi) Open Addresses (a dataset of physical and postal addresses worldwide, including the UK), (xii) the Scottish and Northern Irish energy performance certificate (EPC) register and (xiii) the BT A-Z directory. The Response stated that in addition many addresses had been inferred, giving the example of inferring from the existence of Flat 8 that Flats 1-7 were also present at the same location.

One last identified source emerged from the evidence of the defendants’ expert, Mr Coyne, who identified GoCompare (an insurance website) as a further source.

Gareth Robson, Chief Data Architect at HM Land Registry, estimated that roughly 90% of existing addresses are added to the Land Registry PPD using PAF data. Mr Smith dealt with this by saying in his fifth witness statement dated 7 March 2025 that he had stopped importing Land Registry addresses in August 2024 and that he had carried out a process which probably removed all of these. He said that he would find alternative sources.

Mr Evans explained that EPC Scotland uses PAF data to capture and cleanse its database under licence from RMG. Mr Smith said in his fifth witness statement that he had removed EPC Scotland data from GetAddress. He said that he kept a full record of the changes made. He exhibited a large table of what at first seemed to be a very large number of changed addresses but which on closer inspection displays just one address in Peterhead with many different postcodes.

The evidence of Mr Smith’s own expert, Mr Coyne, was that the method used by Mr Smith to remove Land Registry and EPC Scotland addresses would not have been comprehensive.

Mr Evans said that the Ordnance Survey Code-Point dataset, the NHS postcode directory contain PAF data, in both cases under licence from RMG.

The defendants’ pleaded assertion that any PAF data obtained from these sources was in each case, or in any instance, provided under OGL terms such that RMG had licensed the defendants to use the data was not established at the trial. I conclude from the foregoing that Mr Smith downloaded a substantial amount of PAF data from these sources for incorporation into the GetAddress Database, sufficient for infringement of RMG’s copyright and database right if the GetAddress Database was marketed by the defendants without a licence.

Then there is IDDQD’s GBR Database. As I have said, it was built around the PAF. Mr Smith opened an account with Ideal Postcodes in his own name in late 2015. He signed up for a second account under the name ‘Colin Berry’ on 24 November 2018.

The defendants admitted downloading parts of the GBR Database but said that all they were doing was verifying, validating or cleansing the GetAddress Database, created from other sources, for maintenance.

Even if the assertion were accurate, it is not clear to me that this makes a difference. The verifying, validating or cleansing apparently involved downloading parts of the GBR Database. Subject to a defence of mere consultation, which I will discuss below, ‘verifying’ and ‘validating’ would at the least involve temporary downloading of RMG’s data, infringing both database right and copyright subject to specific defences considered below. ‘Cleansing’ on its face involves removing false data and replacing it with correct data, thereby incorporating the GBR data into GetAddress and making it available to customers. This was confirmed by Mr Smith in cross-examination when he said that he went to Ideal Postcodes to obtain data if he could not find it more cheaply elsewhere and then added that data to GetAddress.

The overall point taken by the defendants in this part of the dispute was that they neither took nor sold a substantial part of the PAF. The issue, in other words, is whether what was taken, either for the initial creation of the GetAddress Database or for its subsequent maintenance (i.e. modification), included the PAF in substantial part.

The claimants’ expert, Mr Beckett, pointed out that two ‘seed’ addresses planted by RMG in their PAF data appear in the GetAddress Database. Mr Smith contended that this must have been the sum total of PAF seeds in the GetAddress Database because seeds are difficult to remove. It was a point he put to Mr Beckett and Mr Evans. They both disagreed and explained why. Seeds have postcodes which do not exist. The maintenance of the GetAddress Database by comparing it with other sources of address data would have flagged up false postcodes. Mr Beckett and Mr Evans expected that most seeds with false postcodes would then have been removed by Mr Smith.

Mr Beckett found a high correlation between PAF data and GetAddress data, in the region of 78%, in certain files compared. He pointed to similarities between the databases in the more frequent use of ‘Bldg Soc’ rather than ‘Building Society’ and ‘C of E’ rather than ‘Church of England’. Mr Beckett also drew attention to many typographical errors in common between the PAF and the GetAddress Database. None of this was challenged.

Mr Beckett reached the conclusion that accordingly there was strong evidence consistent with the GetAddress Database having been obtained directly or indirectly from the PAF. In cross-examination he did not resile from that view.

Mr Coyne was adamant in his written evidence that the defendants had not incorporated any of the PAF into their GetAddress Database. But in cross-examination he accepted that (a) the lack of records kept and disclosed by the defendants going back further than 2024 provided a significant limitation to his ability to give an opinion on the matter, (b) the content of the current GetAddress Database was consistent with the defendants having incorporated part of the PAF and (c) accordingly the court was obliged to look at evidence other than expert evidence in order to reach a conclusion as to whether any of the PAF was taken and if so how much.

Following Mr Blanchard’s suspicions that the defendants were taking IDDQD’s data and using it contrary to the terms of the agreement with IDDQD, he wrote what he called ‘sleeper software’ to gather evidence against the defendants. Whenever Mr Smith’s account made a request for addresses using a particular postcode, the software was able to collect all the addresses for that postcode existing on the GetAddress Database before any data was transferred from the GBR Database, if that was going to happen. After the requests had been concluded, the GetAddress Database was checked again by the software to see whether addresses using that postcode had now been added. In addition, Mr Blanchard introduced errors into some of the addresses requested, which he called ‘sleepers’. Mr Blanchard’s evidence was that on 9 August 2022 Mr Smith downloaded over 70,000 addresses from the GBR Database and Mr Blanchard found them to have been added to the GetAddress Database. Within these, 5,855 sleepers were found. Mr Blanchard maintained this evidence in cross-examination.

The experts were agreed that of the 5,855 sleepers around half were still present on the GetAddress Database two years later in October 2024. They also agreed that 84% of the matched sleepers were added to the GetAddress Database within 2 hours of the search of the GBR Database on 9 August 2022.

Between late 2015 and 13 February 2023 when IDDQD terminated these accounts, according to Mr Blanchard 196,599 postcode searches were made on the two accounts. Mr Blanchard said that on average there are 17-18 addresses which use a single postcode, so this means that the defendants would have extracted 3-4 million addresses from the GBR Database, around 10% of all UK addresses. The defendants’ expert, Mr Coyne, put the figure at up to 7.2%.

Either way, for that reason alone, given that the GBR Database is built around the PAF the defendants must have downloaded PAF data. The PAF data thus downloaded during maintenance of the GetAddress Database would have joined the PAF data used in the creation of the GetAddress Database.

By the time of closing argument the defendants accepted that 5.3% of the GBR Database was taken. Even if correct, and it may well be an underestimate, it is likely that the PAF data within it and the PAF data used in creating the GetAddress Database constituted significant human, technical and/or financial investmenton the part of RMG and thereby made up a substantial part of the PAF.

It follows from the foregoing that Mr Smith extracted a substantial part of the PAF and incorporated it into the GetAddress Database. That substantial part was re-utilised by Codeberry by the marketing of GetAddress. Subject to any defence, the defendants infringed RMG’s database right in the PAF.