Two teams report revenue for the same product line and get different numbers. Not because anyone made an error, but because they're using different definitions of "revenue" — one includes refunds, the other doesn't; one recognises at invoice, the other at payment. This is semantic drift, and it's one of the most common and most damaging data governance problems in enterprise organisations.
Nobody changed the definition on purpose
Semantic drift rarely happens through deliberate change. It happens when a new team builds a dashboard using their own interpretation of a metric. It happens when a business rule changes but only some downstream systems are updated. It happens when an analyst creates a "corrected" version of a KPI that becomes the de facto standard in their department while the official version diverges.
Owned, versioned, enforced
Preventing semantic drift requires three things: clear ownership, versioned definitions, and technical enforcement. Every critical KPI needs a named owner — the person who decides what the metric means, approves changes to its definition, and is accountable for its consistency across all reporting surfaces.
Definitions should be codified in a semantic layer — a single source of truth that all reporting tools query against. Changes to definitions go through a pull request workflow with business stakeholder review. This doesn't slow things down; it prevents the far more expensive problem of discovering six months later that two executive dashboards have been telling different stories.
Thirty KPIs, not three hundred
Don't try to govern every metric at once. Identify the 20-30 KPIs that drive executive decisions and business operations. Get those right first — with clear owners, versioned definitions, and a shared semantic layer. Expand from there as the governance discipline matures and the value becomes undeniable.