Inducements and the Retail Investment Strategy: Instrumenting the New Disclosure and Best-Interest Tests

A best-interest challenge you cannot reconstruct from data is a challenge you lose. The Retail Investment Strategy does not just tighten the conduct rules; it moves the burden of proof onto records your systems may not currently produce.

On 18 December 2025 the Council and the European Parliament reached provisional agreement on the Retail Investment Strategy package, with technical finalisation expected in early 2026. Most of the commentary has been about who won the fight over inducements. That is the wrong thing to watch. The consequential change for a technology function is quieter: the reform converts a set of advice judgements that firms used to summarise in a policy into a set of decisions each firm must be able to reproduce, per client, from data. Treat the numbers below as provisional until the final text is published, and verify article references against your transposing national law.

What actually changed for the adviser

Three things matter for the people who build advice systems. First, a partial inducement ban: a new provision in MiFID II (introduced as Article 24a) prohibits inducements paid by product manufacturers to distributors in relation to the reception and transmission of orders, or their execution, for retail clients where no advice is given. It is not the blanket ban the Commission originally floated; where advice accompanies the transaction, inducements survive, subject to a strengthened test that they deliver a tangible benefit to the client and are disclosed clearly and separately from other costs. Member states remain free to go further and impose a fuller ban.

Second, a sharpened best-interest test. An adviser must base a recommendation on an assessment of an appropriate range of suitable products, recommend the most cost-efficient product within that range, and offer at least one product without features that are not necessary to the client’s objectives. Third, a value-for-money regime that judges cost against peer groupings and supervisory benchmarks, so that the cost comparison sitting behind a recommendation is itself measured against an external reference. I have written separately on the benchmarking data you will have to report; here the point is what those tests demand of your advice records.

The record is the requirement

Read the best-interest test as an engineer and it stops being a principle and becomes a query. “Most cost-efficient within an appropriate range” is only meaningful if you can show, for a given recommendation, which range was considered, what the cost-efficiency comparison returned, and why the recommended product won. Where an adviser recommends something more expensive than an equivalent, the obligation is to justify that objectively and keep the record of the justification. That single clause is the whole engineering brief. A policy that says “we always recommend in the client’s best interest” evidences nothing. A record showing the products screened, the costs compared, the benchmark applied and the reason a higher-cost product was chosen evidences everything.

Most suitability engines were built to reach a decision, not to preserve the reasoning behind it. They score a client’s profile, filter a product universe, surface a recommendation, and print a suitability report. What they frequently do not do is persist the intermediate state: the candidate set before filtering, the cost figures each candidate carried at that moment, the benchmark value in force on that date, and the specific rule that eliminated the cheaper alternative. Once that state is gone, the recommendation is unreconstructable. You can assert it was in the client’s interest; you cannot demonstrate it.

Which decision points to capture

For each advised recommendation, the record a supervisor or a complainant will want to reconstruct contains, at minimum:

• The client profile inputs as they stood at the point of advice, versioned, so a later profile change does not silently rewrite history.
• The appropriate range considered: the product universe that was in scope, and the business-model rationale for that universe.
• The cost-efficiency comparison: the cost and charges figures for each candidate at that moment, and the peer or supervisory benchmark value applied on that date.
• The elimination logic: which rule removed the cheaper or feature-lighter alternatives, and whether a no-unnecessary-features option was presented.
• The objective justification, as free text tied to the above, where the recommended product was not the cheapest.
• The inducement position for that transaction: whether any was received, its amount, the tangible-benefit rationale, and the disclosure served to the client.

None of that is exotic data. All of it exists transiently inside the advice workflow. The failure is almost never that the firm cannot compute it; it is that the firm does not retain it in a form that binds the inputs, the comparison and the outcome into a single, immutable, timestamped record.

Where firms will instrument this wrong

The common failure modes are predictable. Cost data is pulled live at recommendation time and never snapshotted, so when charges change you can no longer show what the client was actually compared against. The benchmark is a moving reference held in a separate system with no point-in-time record, so the comparison cannot be reproduced six months later. The disclosure document is generated and sent but not archived against the transaction, so you can prove a template existed but not that this client received these figures. And the justification field is optional, free-text and unvalidated, so it is either empty or a phrase copied across a book of business, which is worse than empty because it looks like a pattern of unconsidered advice.

These are the same disciplines that turn logs and design documents into defensible evidence elsewhere in the estate. If you have already built a single control fabric for evidencing across other obligations, the advice log belongs in it rather than in a bespoke silo. And the retention model should assume scrutiny: marketing communications carry a five-year record requirement, extendable to seven at a competent authority’s request, and advice records should be held to at least the same horizon, because a disclosure-of-evidence duty can reach records you assumed were long past their useful life.

What a challenge looks like in practice

A retail client complains, two years on, that they were sold an expensive product when a cheaper equivalent existed. The supervisor does not want your policy. They want the record: what range you considered, what the cheaper product cost that day, what benchmark you measured against, and the justification your adviser recorded for going higher. If your systems can return that in a single, coherent, timestamped bundle, the challenge is answerable on the facts. If they can only return a suitability report that states a conclusion and a policy that states an intention, you are defending a decision you can no longer show you made properly.

The Retail Investment Strategy did not just raise the conduct bar. It moved the evidence of clearing it out of the policy and into the record. Build the advice log for the challenge you cannot see coming, because that is the one that arrives.