Single source of truth

Every event team knows the Monday-morning scene: registration platform says 4,812 attendees, the CRM says 5,040, someone's spreadsheet says 4,650, and a board meeting needs one number by noon. A single source of truth ends that argument structurally. You designate, per data type, which system is authoritative — registrations live here, exhibitor contracts live there — and wire everything else to read from it or write back to it. Edits happen in the source system; copies are treated as disposable views. For organizers the payoff compounds: reports reconcile, matchmaking runs on current data, and cross-edition comparisons stop being archaeology. It's also the precondition for AI. Agents and models amplify whatever data you feed them, and an agent reading three conflicting attendee lists will confidently act on the wrong one. Clean, resolved, authoritative data isn't the boring prerequisite to AI — it's most of the work. The honest nuance: "single source" describes authority, not necessarily one giant system. A registration platform can own attendee data while the CRM owns exhibitor relationships; what matters is that ownership is explicit and syncs flow one agreed way. The common failure is the shadow spreadsheet — the export someone "just cleaned up locally" that becomes its own competing truth. Fight it with access, not policy: when the real system is easy to query, the spreadsheets die on their own.