The AI initiative got a name in January. A sponsor. A number.
Before any of that, somebody ran a readiness assessment. It came back saying the data was not ready. They always come back saying that.
So the org did the responsible thing and funded a cleanup. Dedupe the accounts. Standardize the case types. Make the fields required. Find owners for the records that never had one. That work ran most of a year, and it was done well, by people who cared about doing it well.
The data is cleaner than it has ever been.
The initiative is still stuck. And the meeting where someone has to explain why is on your calendar.
Here is what the cleanup fixed.
Every platform got more accurate inside its own walls. Cases in Salesforce are cleaner. Change records in ServiceNow have owners. Jira looks better than it has in years.
Each system is a better version of itself now.
None of them know anything about each other. They never did. Nobody noticed, because until somebody asked a question that crossed all three, it never came up.
Clean and connected are two different things.
Cleaning a record makes the record more correct. It does not attach that record to anything.
Think about what you actually want to know. A change went into your Salesforce org on Tuesday. Wednesday, tickets started. Are those the same event?
Nobody left that field blank. There is no field. It was never required, never optional, never a table anybody built. The relationship is real and it lives in the operation; the systems have no record that it exists.
You cannot backfill a field nobody created.
So the model works with what it has.
Give it a case and it does well. Summarizes it, routes it, tags it, drafts something reasonable back to the customer. That is real work and it is worth having.
Then ask it why the case happened.
Now it needs three systems it has never been introduced to. It does not know what shipped. It cannot see the change record. It has no way to tell that eleven tickets from four accounts are one thing and not eleven.
So it gives you a good summary of a case. Which is what you asked for. Not what you needed.
The org reads this as an AI problem.
Vendor evaluation reopens. Someone suggests a different model. Someone else suggests a different platform. Another six months, another sponsor, another number.
You don’t have an AI adoption problem. You have a correlation problem that predates AI by a decade. AI just showed up recently enough to take the blame.
Here is how you know it is not a model problem.
Somebody in your company can already do this. It is probably you.
Give the right person the four logins and an afternoon and they come back with the answer. Not because their data is better. Because they know which team pushed which change. They know a contractor touched that integration last month, roughly when, and that nobody wrote it down anywhere. They know which of those two things the tickets are actually about.
That knowledge is real and it is not in any system. It is in a person. At most companies it is in one or two of them, and you know exactly who they are. So does everybody else, which is why they get the call.
That is the cost. Not the cleanup budget. The afternoon, over and over, from the only people who can do it, every time something breaks.
The cleanup was never going to touch that. It was not built to.
Teams who cleaned the data, watched the initiative stall anyway, and are now being told the answer is a different model are exactly who we’re thinking about.
If that’s your team, we offer a complimentary Investigation Cost Audit. Forty-five minutes. Structured diagnostic across five dimensions. You leave with a scorecard that quantifies what the investigation is actually costing you: in time, in recurrence, in the correlation work no cleanup project was scoped to do.