Fourteen compliance systems still need an answer about authority

A marketing reviewer searches for guidance on a claim and finds a policy, a previous approval and an email from a specialist. All three are relevant. They may also disagree. The policy could have changed, the earlier approval might concern a different audience and the email may describe an exception that was never intended to become standard practice.

An AI assistant can make those records easier to find and explain. That is useful work, but it leaves the organisation with a question that retrieval alone cannot settle: which record should govern the decision being made now?

MCO’s report, Global Compliance Outlook: Rising Ambition, Fragmented Reality, draws on research conducted by The Harris Poll and reports an average of 13.8 compliance systems among the firms represented in its survey. The figure is a striking account of the technology environment, but the stronger implication concerns authority across that environment. An answer can become faster without becoming better founded.

The survey describes a particular market

The research surveyed 370 senior compliance professionals in the US, Europe and Asia-Pacific. Participants worked in financial services firms with at least 500 employees and held director-level or more senior roles with input into compliance decisions. The results are therefore not a census of every financial services business, and the average should not be applied uncritically to a small lender or an individual marketing function.

MCO is a compliance technology provider, and the release supports its argument for an integrated platform. That commercial context does not invalidate the research. It does mean that the survey and the vendor’s proposed solution should be read separately. The system count cannot establish that a particular architecture will improve decision quality, nor does it show that every specialist application is redundant.

Different systems can exist for good reasons. A business may need dedicated records for employee conduct, transactions, third-party risk or marketing approvals. Combining them into a single interface does not automatically make their concepts equivalent. The useful question is whether the organisation can connect the evidence relevant to a decision and resolve differences between the records it finds.

The source release provides methodology but does not state the survey’s fieldwork dates. That limits how precisely the snapshot can be located in time. It also reinforces the need to describe the reported finding as research among a defined group, rather than a timeless property of the sector.

Fragmented compliance records need a test of authority

Search systems generally reward a close match to the question. A previously approved campaign may contain exactly the wording a marketer wants, making it a strong retrieval result. Its authority depends on something else: the reason it was approved and whether that reason still applies.

Imagine an illustrative product claim approved for a professional audience, with a prominent explanation elsewhere in the same document. A retail campaign using the sentence alone would lose both the audience condition and the supporting context. The wording is an excellent semantic match and a weak precedent. More complete retrieval may expose those differences, but the decision still needs a recognised source of interpretation.

This is why organisational memory needs the limits of expert judgement. A record becomes reusable when the next user can see which conditions made it valid. Without those conditions, historical decisions can accumulate the appearance of authority simply because they are easy to locate.

A practical record should distinguish a current policy from a superseded one, a general interpretation from a case-specific exception and a recommendation from an authorised decision. It should identify who owns the position and where the source evidence sits. Those distinctions matter even if every document is stored in one platform, because consolidation of location is not consolidation of meaning.

Disagreement needs an owner

The hard case is not the missing document. It is the situation in which several apparently credible records support different answers. An assistant may be tempted to combine them into a smooth explanation. That can remove the very disagreement the organisation needs to resolve.

If the latest policy conflicts with a retained approval, the result should expose the conflict and direct it to the appropriate owner. If an exception was granted by someone without authority to alter the wider policy, the system should preserve that scope. If no one can identify the current position, the output should not manufacture one from whichever source appears most often.

These are proposed operating disciplines, not findings measured by the MCO survey. They follow from the problem of using fragmented records as inputs to a consequential decision. The value of AI is partly that it can make conflicts visible earlier, provided the workflow rewards surfacing them rather than delivering a confident answer at any cost.

That also changes what a business should ask of integration. The connection needs to carry status, ownership and applicability as well as content. A document sync that loses its review state may expand access while weakening the reader’s ability to judge the record. A summarisation layer that omits an exception’s expiry condition can turn a narrowly accepted case into an apparent general permission.

Fewer tools is not the only useful outcome

There may be a sound commercial case for consolidating applications, especially where teams duplicate entry or struggle to assemble records. But the number of systems is a poor sole measure of organisational intelligence. A business can have one platform and no reliable way to determine which decision is current. Another can retain specialist systems and maintain a clear account of how their evidence relates.

For a marketing leader, a more revealing test is to follow one disputed claim. Can the team identify the governing position, explain why it applies and locate the person who can reconsider it? Can a later change be communicated to the live work that depends on it? These questions concern the organisation’s ability to act on knowledge rather than its ability to collect it.

The wider marketing test of organisational AI asks whether adoption changes how work moves through the business. Authority is part of that movement. An assistant should not merely bring more information to the reviewer; it should help the reviewer understand which information can carry the decision.

The promise of connected compliance is therefore more demanding than a better search box. It is a business in which relevant evidence can be found, its status understood and unresolved interpretation brought to someone able to decide. The useful reduction is in the distance between information and accountable judgement, whatever the final system count happens to be.