Dictionary and user-defined fields
Five settled columns per field, with custom fields added to your institution’s logic under the same rules.
- Baseline frozen
- Same route for custom fields
Capability · Data governance
Governance works when nobody is back-filling data in filing season
Failure is easy to recognise: filings close only because someone reconciles by hand, a week before the annual report everyone hunts for numbers, and one metric reads three different ways across three levels. We deliver a dictionary with named owners, a tiered metric library, a change process and quality rules — built in 6–12 weeks, reviewed quarterly afterwards.
Position
A definition is a management decision that has been written down
Every field in the dictionary carries four mandatory entries: source, calculation, refresh frequency and owner. Miss one and the field is just a column of data; keep all four and it becomes something a reviewer can challenge. That is why governance is signed by business, risk and finance, never by a system administrator on their behalf.
The second commitment is customisability: dictionary definition and user-defined fields run through one mechanism. New fields follow the same change process, so a policy update does not wait for a product release, and nobody opens a private spreadsheet to bypass the dictionary.
An unowned data dictionary is more dangerous than none: it lets every department say governance has been handled, while the back-filling in filing season continues unchanged.Catalpa delivery method
Six moves: settle the conflict before writing the rule
The order is fixed. A change process that arrives before the dictionary is signed will govern fields nobody has agreed on.
Inventory the definitions in use
Compare live reports, filing templates and system fields side by side, recording every case where one metric has two calculations. Output: conflict register ranked by filing impact.
Sign the dictionary and name field owners
Source, calculation, refresh and owner per field; three signatures; baseline version frozen. Output: data dictionary and owner matrix.
Fix the definition-change process
Five settled steps: request, impact analysis (which reports, models and disclosures move), version number, effective date, notification scope. New and amended fields follow the same route.
Build the metric library in three tiers
Atomic metrics read straight from fields, derived metrics combine them by formula, disclosure metrics serve filings and LP reporting. Each tier references the dictionary instead of restating it.
Turn on data quality rules
Completeness, consistency, timeliness and permission checks run automatically; exceptions open a task addressed to the field owner, not an entry in a reminder list.
Issue the consistency audit
Sample the same metric across three levels, filings and disclosure documents; state the difference and the remediation owner line by line. Output: definition-consistency report, the acceptance artefact.
Artefacts, duration, acceptance
6–12 weeks is the standard range from agreement of the work-breakdown structure. Hierarchies with many filing templates phase against the roadmap and share milestones with platform delivery.
| Artefact | Typical duration | Acceptance method |
|---|---|---|
| Definition-conflict register | Within 6–12 weeks | Three lines confirm nothing material is missing |
| Data dictionary and field-owner matrix | Within 6–12 weeks | Three signatures; exactly one owner per field |
| Change process with versioning rules | Within 6–12 weeks | One live change run end to end in the review meeting |
| Metric library: atomic, derived, disclosure | Within 6–12 weeks | Formulas spot-checked against the dictionary; disclosure tier traceable |
| Quality rules and exception tasks | Within 6–12 weeks | Seeded bad data triggers rules and reaches the owner |
| Definition-consistency audit | At close | Zero-difference rate and remediation closure listed line by line |
| Quarterly definition review | After acceptance | Recalculated to the same formula, compared with prior period |
Reuse
The 18 indicators and valuation logic need building once
Portfolio risk monitoring, KPI alerts, asset valuation and exit models consume the same derived metrics. Where an in-house team exists, we hand this batch over as reusable components tied to the dictionary rather than letting two implementations diverge.
- Indicators recalculated daily
- Valuation and disclosure share source
- Formula versions traceable
Governance base
Six mechanisms behind the dictionary
The dictionary is the start. Without these six, definitions drift between the next two policy changes.
Definition change control
Request, impact analysis, version, effective date, notification. Affected reports and disclosures are listed automatically.
- Impact visible
- Changes logged
Three-tier metric library
Atomic, derived and disclosure tiers maintained separately; the disclosure tier cites a derived formula rather than inventing one.
- Reference, not restatement
- No duplicate definitions
Valuation and disclosure metrics
Valuation, share and return bases, LP reports and external disclosure draw on the same fields, so two ledgers cannot form.
- One valuation basis
- Traceable disclosure
Data quality rules
Completeness, consistency, timeliness and permission checks run on schedule; exceptions become owner tasks.
- Four check families
- Exceptions as tasks
Single filing outlet
External filings and LP/GP disclosure generate from one outlet, so a metric is not computed twice by two departments.
- One outlet
- Versioned templates
Object domains under governance
Scope is cut to your business lines. Anything excluded is marked as not covered in the audit rather than assumed.
Funds and share classesProjects and transactionsPortfolio company entitiesOwnership layers and dilutionValuation and impairmentRisk indicators (18)LP reports and disclosureSupervisory filing templatesEffort and costSub-fund reporting
Who owns it
Responsibility does not move with hosting mode
Field owners and change approvals stay with the client in public cloud, private cloud and intranet alike; we supply the process tooling, the review and the quarterly audit. On intranet, comparisons and audits run inside your environment and dictionary changes are logged as your locality requires.
- Owners are client-side
- We audit and review
- Nothing leaves the intranet
Who reads it
Three audiences consume the same definition
A supervisory template, an LP question and an auditor chasing a formula each want a different cut of one definition. Where the capital comes from decides where conflicts arise: hierarchies conflict across levels, market managers across valuation and exit models.
- Filing definitions
- Disclosure definitions
- Audit trail
Published method
Why definitions come first, in print
Definitions before features, why filing season needs manual back-filling, why implementation schedules compress — these arguments are published, so they can be quoted straight into an approval paper.
Read the articleWho should not buy this
Anyone wanting a standalone dictionary template: without signatories and a change process it is stationery, and we do not sell it. Institutions that will not name field owners are equally unsuitable — governance ends with a person answering a challenge, and an unowned dictionary gets routed around within six months.
A single-fund team that will not run any system and collaborates in spreadsheets should consolidate the files first. That is cheaper than a dictionary.
Questions governance leads ask
Answers match our acceptance method and are quotable internally.
How is a data dictionary different from a field specification?
A field specification says what the database holds. A dictionary says how the business computes it, how often it refreshes and who owns it. The signatories are business, risk and finance, not development — which is what makes it usable for filings and disclosure.
How long does governance take, and what sets 6–12 weeks?
Hierarchy depth, number of filing templates and how exportable current systems are. Six to twelve weeks is the standard range from agreement of the work-breakdown structure; fields nobody will sign off are listed as open items instead of stretching the schedule.
Governance first, or platform first?
Depends on where the pressure comes from. With a hard supervisory deadline we usually govern that one outlet and its fields, then phase the workflows. Where everything is manual, a minimum platform slice goes in first with the dictionary built alongside. Both orders get written into the roadmap; it is not a position of faith.
Does changing a definition break prior disclosures?
Changes carry a version number and effective date, and historical conclusions keep the formula in force at the time. The audit lists differences caused by the change and whether the next disclosure has to explain them.
We have an in-house system. Do we still need a metric library?
Usually the need runs the other way: the implementations exist, scattered across code and reports. We consolidate them into three tiers with owners, so valuation, alerts and disclosure stop computing separately.
What if quality rules generate too many false positives?
Rules are staged. High-impact checks — missing key fields, cross-table inconsistency — open tasks first; the rest run in observation mode. False positives go into an adjustment list and thresholds are revised through the change process.
Next step
Bring three reports and we will collide the definitions
We compare how one metric is computed in business, risk and finance on the spot, separate the fields that need a signature from those that can wait, and place you in the 6–12 week range.