[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"article-body-you-have-enough-data":3},"\nThere is a sentence that ends more AI initiatives than any budget\ncommittee, and it is spoken in a tone of great responsibility. Our\ndata is not ready. The room nods, a data programme is commissioned,\nand the AI conversation is adjourned for two years. Nobody has ever\nbeen fired for saying it, which is part of the problem.\n\nThe sentence is half right. Garbage in, garbage out is not a myth,\nand executives know how their estate would perform under\nexamination. Only around one in nine say their data quality is\nsufficient for AI use ([the verification\ngap](/insights/the-verification-gap)), and audit-detected AI errors\nreaching boards suggest the other eight are correct to worry.\n\nThe half that is wrong is the sequencing, and the sequencing is\nwhere the years go.\n\n## The half that is right, stated precisely\n\nStart with what the readiness instinct gets right, because the\nargument only works if it is honest. Models trained on drifted,\nduplicated, mislabelled data produce confident nonsense, and\nconfident nonsense in front of a decision maker is worse than no\nmodel at all. The caution is earned.\n\nBut look at what the caution is actually about. It is almost never\nabout volume. Two decades of ERP, CRM, incident registers, contracts\nand transaction history sit in the systems of record; the registers\nalone are training data that most functions have never used as such.\nEnterprises are not short of data. What they are short of is\ngovernance where the data gets used: one resolved identity per real\nthing, lineage that survives an export, policies that are enforced\nrather than published, and permissions that hold at the point of\nconsumption.\n\n*You do not have a data shortage. You have enough data, ungoverned\nat the point of decision.*\n\nThat distinction is the first link in a chain. Ungoverned data makes\nthe AI built on it untrustworthy, and untrusted AI leaves decisions\nexactly where they were. What looks like three problems, data, AI\nand decisions, is one chain, and it breaks in the same place it can\nbe fixed. Every year spent waiting adds to the decision debt: the\ncompounding cost of decisions made late, blind or not at all while\nthe estate was being made ready.\n\n## Why the clean-up programme never converges\n\nThe readiness instinct produces a specific project shape, and every\nlarge organisation has run one. Scope the whole estate. Define\nquality in the abstract, \"fit for purpose\" with no purpose named.\nCleanse, deduplicate, catalogue. Declare partial victory at the\nbudget review and quietly relapse the following year.\n\nThe shape fails for a structural reason, not a competence one. A\ndata programme with no consuming decision has no forcing function.\nNothing pulls on the data, so nothing establishes which defects\nmatter, which of the four customer records is canonical, or what\n\"done\" would even mean. Quality is unmeasurable in the abstract\nbecause quality is a relationship between data and a use, and the\nuse was postponed to phase two. The programme cannot converge on a\ntarget that has not been named.\n\nMeanwhile the estate keeps moving. Systems are added, fields drift,\nand yesterday's cleansed table starts decaying the day the project\nteam leaves. A one-time clean of a continuously drifting estate has\nthe same half-life as any one-time fix of a living system, which is\nto say, one budget cycle.\n\n## Invert the sequence\n\nThe alternative is not to skip governance. It is to reverse the\norder and let the decision drive it.\n\nPick the decision first. The decision names its data, which is\nalways a fraction of the estate; a demand forecast does not care\nthat the marketing taxonomy is a mess. Wire the decision up, and\nevery defect that matters now surfaces at the exact point it bites,\nin a prediction that misses, an exception that misfires, an audit\ntrail that will not reconcile, with a named owner already attached\nbecause the decision has one. The forcing function the clean-up\nprogramme never had is simply use.\n\nRun this way, governance stops being a programme and becomes a\nconsequence of use. Entity resolution happens because the decision needs one\nsupplier, not four spellings. Lineage exists because every figure\nmust show its account of itself. Policy conformance is watched\ncontinuously because breaches now have somewhere to land. The data\ngets clean in the order the business actually needs it clean, and\nit stays clean because something is pulling on it every day.\n\n*The decision layer is the forcing function. Data quality is its\nby-product.*\n\nThis is also where the readiness excuse quietly inverts. Waiting\nfor clean data before wiring decisions guarantees neither. Wiring\nthe decision first delivers both, and the organisations doing it\nare not braver, they are sequenced correctly.\n\nProphesee is built around that inversion, and Prophesee Data\nGovernance is where the by-product becomes visible. An ontology resolves the corpus\ninto entities, owners and applicable policies. The policy is\nwatched continuously, with every breach routed to a named owner.\nGoverned agents execute the approved clean-up, inheriting\npermissions, budgets and audit by construction. Governance produced\nthis way is hard to copy for the same reason it is hard to fake; it\nis generated by operating decisions, not written beside them.\n[Start here](/contact).\n",1786833838954]