I am an academic researcher constructing a board-interlock network of UK companies for a non-commercial academic study. I would like to request the officers bulk data product (or the closest available extract) covering the following:
Data points needed, per officer record:
- Company number (to link to Companies House/FAME identifiers)
- Officer/person unique identifier (essential — needed to match the same individual across multiple company appointments; without this, interlocks cannot be detected)
- Officer full name (as a fallback match key alongside the unique identifier)
- Officer role/type (director, secretary, LLP member, corporate officer, etc. — so I can restrict to natural-person directors)
- Date of appointment
- Date of resignation/termination (or confirmation the appointment is still active)
- Officer status (active/resigned)
- Nationality and country of residence (if included in the standard extract — not essential, but useful if available at no extra cost)
Coverage requested:
- All UK-registered companies (not restricted to a sub-sample), covering appointments active or occurring at any point between 2008 and 2023 (the earlier years are needed to establish pre-existing network ties before my main 2018–2023 study window)
- Both the historical snapshot and, if available, the daily/periodic update files for that window, so appointment and resignation timing can be reconstructed accurately over time
Format: Whatever your standard bulk delivery format is (I understand this is typically a fixed-width text file with an accompanying data specification) is fine — I can parse it on my end.
Purpose: This data will be used solely to construct board-interlock network measures (e.g., shared directorships between firms) for a peer-reviewed academic paper on corporate governance and UK companies. No commercial use is intended.
Please let me know if there is a standard request form or additional information you need from me (institutional affiliation, etc.) to process this.
Sincerely,
Mostak Ahamed, PhD
@mostak84 This is actually a far more involved process than a simple parse unfortunately. I built a director matcher across companies for convert-ixbrl.co.uk while building an IXBRL to JSON API and it was anything but straightforward.
Key issues being that person ID is not actually unique, same individual can get different IDs.
Using other information to link one person across companies is not easy either as CH only publish month and year of birth, addresses are not always up to date, variations in name spellings or order to name a few challenges.
As this is a research project and you seem to be technically inclined, I’d be happy to outline exactly the steps I took so you can replicate this at your end without going through the same trial and error process I did. Feel free to dm me.
Hi Mostak,
Your board-interlock research sounds very interesting.
I’ve been developing a Director and Company Intelligence Portal that analyses relationships between companies and directors using Companies House data.
One of the most valuable aspects has been the ability to examine director appointment histories across multiple companies and then navigate from a director to all associated organisations. This makes it possible to explore company-director networks, historical appointments, and connections between businesses through shared directors.
Product 195 has provided an excellent foundation for the company side of the analysis, particularly for company search, profiling, and linking director activity back to company records.
I can certainly see how a bulk officers dataset, combined with the company data, would support research into board interlocks, governance networks, and longitudinal relationship analysis.
I’d be interested to hear how your research progresses.
Regards, Stewart Palmer
Hi @effahid, thanks for offering to share your experience. I’ve been tackling similar identity-matching challenges in my research project, with these results (this data represents the snapshot for August 2026):
| Evidence |
Source observations |
Distinct identities after resolution |
| Individual Officers |
45,583,116 |
21,632,557 people |
| Individual PSCs |
13,975,329 |
10,062,527 people |
| Officers and PSCs combined |
59,558,445 |
22,610,788 people |
| Addresses |
73,785,851 |
18,278,644 addresses |
The combined person total accounts for people appearing as both Officers and PSCs.
Do you have any comparable totals or validation results you’d be comfortable sharing? I appreciate that different snapshots, coverage and matching rules will affect the figures, but a rough comparison would be very helpful.
@companieshousedata happy to share my stats for convert-ixbrl.co.uk director monitoring feature. It’s free by the way so you can just try it out to compare a set of directors with yours and to see if the other appointments it finds for a given PSC/director matches with yours.
The snapshot I used for this is from mid-October 2026, so it is a couple of months later than yours.
| Evidence |
Source observations |
Distinct identities after resolution |
| Individual officers |
50,486,027 |
19,940,967 people |
| Individual PSCs |
12,516,944 |
8,273,359 people |
| Officers and PSCs combined |
63,002,971 |
20,636,098 people |
| Addresses (officers and PSCs, all record types) |
73,643,738 |
14,472,275 distinct |
How I counted this
- Officers: Individual appointments only (corporate officers excluded), current and resigned.
- PSCs: Individual persons with significant control only (corporate and legal-person PSCs and PSC statements excluded).
- Identity key: Lower-cased forenames (including middle names) + surname + month/year of birth. Companies House only publishes month and year of birth, so that is as precise as a date-based key can get reallly
- Combined: The officer and PSC keys are put through the same normalisation and de-duplicated together, so about 7.6 million people who are both an officer and a PSC are counted once.
- Addresses: Every officer and PSC record (including corporate ones), keyed on address line 1 + postcode, lower-cased with spaces removed from the postcode. This is a simple exact key, not a geocoded or fuzzy match, so I think there is a possibility of over-count compared with a properly resolved address list.
A few other comments
- The name + month/year-of-birth key gives a figure in the same range as yours for officers (19.9M vs 21.6M), which is reassuring given how different the pipelines are. I’d expect your figure to be a bit higher if you split on address or treat name variants differently.
- Adding postcode to the key for PSCs (name + month/year of birth + postcode) gives 10.05M, almost exactly your 10.06M. Without postcode it is 8.3M, so common name and birth-month collisions matter more than I expected among PSCs.
- Companies House’s own person numbers are not a reliable identity on their own: counting distinct person numbers across individual officers gives 27.2M, well above any name-and-date-of-birth based count, because one real person can carry several of them.
For monitoring individuals over time we use a stricter rule than the counts above: a person is matched only when at least three of four fields agree (forenames, surname, date of birth, address). That trades some recall for very few false positives, which suits alerting better than a headline count.
Comparing the tables, our figures sit within about 10% of yours on most lines. That’s reassuring given how different the matching rules are. Address observations agree to within 0.2% (73.6M vs 73.8M), so we’re clearly counting the same population. Larger gaps I think come from the key selection.
Hope this helps
2 Likes
Thanks very much @effahid, that’s incredibly useful, and I really appreciate you taking the time to provide both the figures and the detail behind how you arrived at them.
I’m particularly reassured that I’m on the right track with my merging rules!
Thanks again!