Your data problem has an owner problem

Three answers. One client.

Picture an operations manager preparing a review pack. The CRM has one address. A provider statement has another. The latest meeting note says the client moved, but does not record when. Every system is working. The pack is still wrong.

The connector is not necessarily broken. Moving information between applications is a technical task; deciding which information should govern the next action is an operating decision. Completing the first can quietly make the second harder.

Our view: an advice firm's first data project should settle a small number of authority questions before attempting a grand unification. Start with the information that determines whether work can proceed.

Choose authority at field level

A single system need not be authoritative for everything. The provider may supply the recorded policy valuation. The adviser may confirm a client's intended retirement date. Operations may own whether the documents required for a particular stage have arrived.

Write those responsibilities down against individual fields or closely related groups. Specify the source, the person who resolves conflicts and the evidence needed to change a value. Avoid giving every department a general instruction to improve data quality. That is how a shared priority becomes nobody's job.

Dates deserve special treatment. A value needs both its effective date and, where useful, the date the firm received it. A statement uploaded today may describe a balance from last quarter. Treating upload time as valuation time creates confidence without freshness.

Keep disagreement visible

Suppose two records disagree about employment status. Do not automatically choose whichever arrived last. A newer import might contain older information. Preserve both observations, identify their sources and send the conflict to someone authorised to resolve it.

The practical output is an exception queue. Each exception needs an owner, a reason it matters and a next action. Prioritise a conflict blocking an imminent client decision above an unused field in an archived record.

The Government Data Quality Framework distinguishes completeness from accuracy. That distinction travels well into advice operations: a filled-in field can still be wrong. The framework is public-sector guidance, not a prescribed architecture for advice firms.

Measure the cost of uncertainty

For a short baseline period, record how often staff stop work to confirm a fact, where they go and whether they get an answer. Count repeat interruptions as well as distinct bad records. One unresolved ownership rule can create the same question across many cases.

The FCA's Consumer Duty board-report findings identify data quality as a weakness when firms try to substantiate conclusions about customer outcomes. They do not prescribe a particular database. The useful management question is whether evidence survives the journey from client work to board assurance.

Choose one workflow and publish a small weekly reconciliation: unresolved conflicts, age of the oldest material conflict, cases delayed and repeat causes. These are proposed management measures, not regulatory reporting requirements.

Buy the connector second

Once authority is clear, integration has a specific job. It can carry approved facts, attach provenance and flag disagreements. Acceptance testing becomes concrete: does the right value arrive, can someone explain its origin, and does a contradiction stop the relevant action?

Start there. A unified dashboard is pleasant. A team that no longer needs to ask which record to trust has changed how the firm works.