8 in 10 Companies Say Bad Data Is Blocking Agentic AI. Outreach Says the Problem Predates the Agents.

McKinsey research cited by Outreach puts a number on a problem sales orgs have lived with for years: eight in ten companies identify data limitations as a barrier to scaling agentic AI. Outreach’s argument is that the barrier isn’t new, AI just makes it visible faster — live CRM pipeline data drifts out of sync with periodic finance exports, reps apply inconsistent deal-stage definitions, verbal-commit and signed-contract get counted as revenue interchangeably, and no one owns data-quality as a job. An AI agent trained on that mess doesn’t fix it; it inherits it, and propagates the errors at scale faster than a human ever could. The upside Outreach reports for customers who’ve fixed the foundation first: AI agents saving 15–21 minutes daily on CRM updates and meeting summaries — modest, but compounding across a full sales floor.

The dollar cost of skipping that step was already quantified a week earlier in the same publication, which found 46% of RevOps leaders still run mostly manual, non-automated processes, 38% cite data accuracy as a top challenge, and 92% of CFOs say accurate forecasting is difficult specifically because of it. The knock-on effects are concrete: only 29% of seller time is spent actually selling, 79% of sales orgs miss forecasts by more than 10%, and 61% of B2B revenue now comes from renewals and expansion — meaning the forecast errors compound into the accounts companies most need to get right.

Put the two pieces together and the sequencing argument writes itself: an org that can’t forecast reliably today isn’t going to get a trustworthy agent recommendation tomorrow by adding AI on top of the same disconnected stack. The 15–21 minutes a day of saved agent time is the reward for fixing data ownership first — not a substitute for it.