DORA · Register of Information · ITS (EU) 2024/2956

A register accepted last year may be rejected this year.

Those are the CSSF's words, February 2026.1 The validation checks did not change — they now apply to more fields. RegisterProof runs your register against the published ESA rules and hands back the rejection list your competent authority would send you, keyed to the template, the column and the row.

First 10 entities · 3 reviews left this cycle

No system access. No OAuth. Deleted on request.

Sample output is illustrative. Rule IDs follow the ESAs' DOR_#### convention; the templates, columns and failure categories are the ones the ESAs name.45

RoI_2026-12-31 / validation.log running…
Running published ESA data-quality checks… Get the fix list
40%
40 of every 100
of entities required to file with the CSSF had done so, two weeks before the 2026 deadline2
CSSF · 17 Mar 2026 · Luxembourg
6.5%
6.5 of every 100 passed
of processed registers passed all 116 data-quality checks in the ESAs' dry run4
EBA · dry run results · 2024
86%
86 of every 100 errors
of all data errors had one cause: a mandatory field left empty5
ESA_2024_35 §30
§ 01
Failure anatomy

Your register does not fail on judgment.
It fails on empty cells.

The ESAs published the breakdown.5 Two templates account for the overwhelming majority of missing mandatory information — and within them, a handful of named columns.

Source
ESA_2024_35
§§ 30–33
TemplateWhere it goes wrongShare of missing
B_02.02 Contractual arrangements — specific information. provider code type c0030 · provider code c0040 · function identifier c0050 · ICT service type c0060 60%5
B_05.01 ICT third-party service provider. head-office country c0050 · ultimate parent code c0080 · parent code type c0090 17%+5
others Spread across the remaining 13 templates. the rest
The columns the ESAs named · select one

Identifiers are the second front

LEI validity for your own entity is checked against the GLEIF register.4 Provider and parent codes matter more than they look: they drive the critical ICT third-party provider designation,3 which is why the ESAs singled them out.5

None of it is interpretive

Every one of these is a deterministic check against a published rule. Worst register in the dry run: close to 50,000 failing data points. The most distinct checks failed by a single entity was 43 of 116.6

§ 02
Since the dry run

The register stopped being a form and became supervisory input

The 2024 dry run was voluntary and best-effort. Two mandatory cycles have run since, the ESAs have used the data, and the bar has moved.

All entries
primary
sources
— until the next reference date, 31 December 2026. Whatever your register says that day is what you file. Your submission window opens after it, and your authority sets the date.9
“A register of information which was accepted last year, may be rejected this year.” CSSF, February 2026 — the checks did not change; they now apply to more data fields
  1. Dec 2024 The ESAs publish the dry-run findings: 6.5% of 947 processed registers pass all 116 data-quality checks; 86% of errors are a mandatory field left empty. ESAs · ESA_2024_35
  2. Apr 2025 First mandatory cycle. Competent authorities report their registers to the ESAs by 30 April 2025. ESAs · collection timeline
  3. Nov 2025 The ESAs designate the first critical ICT third-party providers. “First, the ESAs collected data from the Registers of Information maintained by financial entities.” Your provider and parent codes are what that designation runs on. EBA · EIOPA · ESMA · 18 Nov 2025
  4. Feb 2026 The CSSF opens the window and warns that the same validation checks now run against more data fields — and that last year's acceptance is not a defence. Scope widens to third-country branches. CSSF · 11 Feb 2026
  5. Mar 2026 Two weeks before the deadline, only 40% of entities required to file with the CSSF had done so. The regulator urges entities to submit early “so that potential issues can be addressed in a timely manner.” CSSF · 17 Mar 2026 · Luxembourg figure, not EU-wide
  6. Next cycle Registers are reported by competent authorities to the ESAs by 31 March of each calendar year. Your own window opens earlier and is not the same everywhere — in 2026 the CSSF's entity deadline was 31 March; DNB's was 20 March. Whether you file in full this cycle is also a question for your authority.108 CSSF · 11 Feb 2026 · DNB · 3 Feb 2026
§ 03
Method

Find out before your NCA does

Three steps. No integration, no access to your systems, nothing for your security team to review.

Turnaround
5 business days
  1. Send the register you already have

    Excel, CSV, or a full xBRL-CSV report package. If your register is still a spreadsheet a consultant handed you, that is exactly the input we want.

  2. We run the published checks

    Structural and naming checks first — the layer that discarded 92 submissions outright4 — then the data-quality rules, LEI validity against GLEIF, and the relational keys between templates.

  3. You get a fix list, not a score

    Every finding names the template, the column, the row and the rule it breaks. Forward it to whoever owns that contract and it closes.

§ 04
The offer

A dry run of your register, by hand, before you file

You send the register you intend to submit. You get back every finding the published ESA checks would raise.

Scope
1 register
1 entity
1 cycle
€750 one-off Credited in full against year one
Zero-risk
Refunded if it tells you nothing new. You decide, not us. No conversation, no questions.
  • 01Findings in 5 business days, each naming the template, the column, the row and the rule it breaks.
  • 02Structural checks first — naming and packaging, the layer that discarded 92 of 1,039 submissions before a single value was read.4
  • 03LEI and identifier validation for your entity and every ICT provider, including the parent-undertaking codes the ESAs flagged as most often missing.5
  • 04Relational keys across templates — the references between B_02.02, B_05.01 and B_06.01 that must resolve before anything else is judged.
  • 05Any format. Excel, CSV, or a full xBRL-CSV package.

No system access. No OAuth. Deleted on request.

Redact anything you like — commercial terms, spend, contract references. The checks run on identifiers, codes and completeness, not on what you pay your providers. Your file goes to no one else and is deleted on request.

§ 05
Alternatives

Three ways to learn your register is wrong

Only one of them happens while you can still do something about it.

RegisterProof Submit and wait Advisory engagement
When you find out Whenever you ask, as many times as you want After your NCA validates, inside a fixed window9 After scoping, procurement and a kickoff
What you get Template, column, row and rule ID for every finding A feedback file, and a resubmission to turn around A populated register, and the same problem next year
Cost to try €750, credited to year one. One file, no onboarding Your remaining days before the deadline A budget line you defend before anyone reads a cell
§ 06
In scope

Everyone in this table files a register

The ESAs have not published a breakdown by entity type for either mandatory cycle. This is the most recent one that exists — and it is from the 2024 dry run. It is here for scope, not for currency.7

2024 dry run · voluntary · the only entity-type breakdown the ESAs have published
Type of financial entitySubmissionsData pointsError rate
Credit institutions2602,931,7671.9%
Insurance & reinsurance undertakings2252,935,5793.3%
Investment firms122248,9882.4%
Asset management companies86209,3061.0%
Payment institutions5063,3183.0%
Managers of alternative investment funds4090,0161.2%
Institutions for occupational retirement provision36344,2833.0%
Electronic money institutions3042,3202.5%
Insurance & reinsurance intermediaries1749,9062.8%
Central securities depositories738,1501.2%

Ten entity types with the highest number of data points submitted.7 The dry run was voluntary and best-effort, so these are not a census of entities in scope and the error rates are not a verdict on any sector today. We publish them because they are the only figures of their kind an authority has put out — and we would rather date them plainly than dress up something newer that nobody published.

§ 07
References

Every number on this page, and where it came from

We publish no regulatory value we have not verified at the source. Where a value depends on your competent authority, we say so rather than guess.

10 entries
8 verified
2 deferred
  1. Verified“A register of information which was accepted last year, may be rejected this year” · the checks are unchanged but apply to more data fields · 2026 reference date 31 December 2025
  2. Verified“As of 16 March, only 40% of financial entities required to submit their register of information to the CSSF have done so”
    CSSF, DORA — Register of Information collection — Update, 17 March 2026. This is a Luxembourg figure. We have not found an equivalent EU-wide number published by any authority, so we do not present one.
  3. VerifiedThe ESAs designated the first critical ICT third-party providers using register data: “First, the ESAs collected data from the Registers of Information maintained by financial entities”
  4. Verified6.5% passed all checks · 93.5% had at least one failure · 116 checks · 1,039 received, 92 discarded, 947 processed
  5. Verified86% of errors were missing mandatory information · B_02.02 accounted for 60% of it · B_05.01 for over 17%
    ESAs, ESA_2024_35, §§30–33
  6. VerifiedWorst register ~50,000 failing data points · most distinct checks failed by one entity: 43 of 116 · 235,000+ failed checks over 9,275,000 data points
    ESAs, ESA_2024_35, §§26–28
  7. VerifiedThe register templates, fields and submission format
    Commission Implementing Regulation (EU) 2024/2956, and the EBA's DORA reporting preparation materials, which publish the applicable validation rules
  8. VerifiedEntity deadlines are not uniform: for the 2026 cycle the CSSF's was 31 March and DNB's was 20 March. DNB also accepts a standardised Excel template it converts to xBRL-CSV.
    De Nederlandsche Bank, Reporting DORA registers of information in March 2026, 3 February 2026, read against the CSSF's window
  9. Confirm with your NCAYour submission window and channel
    Entity-level deadlines, portals and packaging are set by each competent authority and differ across member states. We will not print a date for your jurisdiction that we have not read on your NCA's own page.
  10. Confirm with your NCAWhether you file in full this cycle
    Some authorities have applied proportionality between cycles, with reduced or confirmatory submissions for entities without material change. Check your NCA's guidance for the current reference date.
§ 08
Who is doing this

A person reads your register. That is the point.

Concierge
review

Bitgen Labs
Tier2Flow
ForwardCerts
Nothing to onboard
No connector, no OAuth scope, no read access to a production system. One file, reviewed, returned. There is no integration for your security team to assess because there is no integration.
Nothing retained
Your register is read, the findings are written, and the file is deleted on request. It is never pooled, never used to train anything, never shown to another client.
Nobody else in the room
No junior staffed to your engagement, no offshore processing centre. One reviewer, the published ESA rules, your file — and a findings list you can check line by line.

I am Jerusan. I build small, exact software for regulatory filings — a Tier II EPCRA filing service and an ACORD 25 certificate checker are the other two. RegisterProof is the third, and it starts the way the other two did: by hand, on real files, before a line of product code.

That is deliberate, and I will not dress it up. The validation rules are published; running them against your register carefully is the whole job, and doing it by hand first is how I find out which parts are worth automating. The software follows the demand, not the other way round. You are buying the findings, and the findings are real either way.

Refunded if it tells you nothing new Credited against year one jerusan@bitgenlabs.com