Make the connection explain itself

Participants receive an invented set of party pairs with shared contacts and different supporting circumstances. They must decide which relationships the evidence supports, which remain uncertain and what further examination would be proportionate. The exercise begins before a matching percentage is shown, so that the group has to explain the meaning of the proposed connection rather than rely on a preferred score.

The discussion then introduces the fictional 920-pair review and the decision to withdraw an automatic merge. Groups examine the delivery sponsor's argument for simplifying the register, the continuing work created by unresolved pairs and the difference between correcting a proposal and claiming a live incident. They prepare a short recommendation for an approving forum that keeps those consequences visible.

The session uses synthetic material throughout. It does not invite participants to submit customer files or treat the exercise as an identity-verification service. The intended result is a more precise conversation about evidence and handling responsibility. An actual workshop would require an agreed purpose, suitable audience preparation and a scope that respects the organisation's professional and information-handling arrangements.

A fictional policy reference contains a contracting party, a contact and a financing relationship. The receiving system offers one familiar category. Participants examine what a user might infer from that category and whether the source supports the inference. The work is to preserve a usable explanation of the relationship, rather than reproduce every source field or choose whichever label makes the screen easiest to complete.

The 360-policy demonstration provides a concrete disagreement between a concise reporting view and the need to retain role distinctions. Teams identify who should assess meaning, what remains unresolved and how a corrected reference differs from an amendment to an insurance contract. They also explain the five-week delay and the cost of the reconstruction without inventing payment consequences that the case does not establish.

This is a discussion format for leadership judgement, not a source of legal or coverage interpretation. All records are invented. The organiser and responsible specialists would agree the actual boundaries in advance. Participants leave with a short account of the decision a mapping is allowed to support and the questions that must remain with the appropriate owners.

The session presents two matching rules evaluated on the same invented thousand-pair set. One has fewer false links but misses more genuine pairs. Participants calculate both accounts, then explain how the consequences differ for a service lookup, a reporting connection and a decision that could affect a customer. The purpose is to prevent a single favourable percentage from substituting for a clear operating recommendation.

The group considers the assisted pilot proposed in the demonstration, including the 390 pairs not linked automatically under the stricter rule. It asks what further evidence or service route those outputs require and why an unlinked result is not necessarily a different-person finding. A second discussion examines what the sample can support and what must still be tested before a wider use.

No production-performance promise is made. The exercise does not choose a universal threshold or authorise an insurance action. An actual engagement would use synthetic or properly authorised material, with relevant operating owners present. The intended output is a recommendation that names both errors, their consequences and the work required outside the automatic path.

Participants begin with a sound identity link and a proposed new use. They must identify what additional evidence is needed before the person receives the role assumed by the workflow. The distinction is deliberately concrete: locating the right person does not establish that person's authority to act for another party or inspect every connected record.

The fictional portal case supplies 240 proposed connections, with 196 supported roles, forty-four requiring review and fifteen ultimately withheld. Groups explain the limited roles clarified for twenty-nine cases and the assisted route retained for unresolved service needs. They also examine the delivery sponsor's concern that repeated review could delay adoption, then prepare an account of the narrower recommendation and its six-week cost in time.

The exercise is designed to improve coordination between the people who maintain relationships and those who authorise their use. It does not inspect a live permission system or offer a security certification. All examples are synthetic. The organiser would agree the scope and involve relevant operating, legal, privacy and technology owners before adapting the discussion to a real enterprise.