Somebody signs off on standardization with a clean two-year plan. Pick the winning tool, migrate everything onto it, retire the rest, book the savings. Two years later the deck gets refreshed, the timeline slides right, and the same slide shows up in next year's planning cycle wearing a new date.
This is not a failure of execution. It's what standardization does when you run it against a real enterprise. The plan assumes convergence is a project with an end. In practice it's a direction you can push in without ever arriving, and the reasons are structural, not managerial.
01The content nobody will migrate
Every estate has a long tail of reports that are used, load-bearing, and owned by someone who will absolutely escalate if they break. Migrating them means rebuilding logic that lives in one analyst's head, revalidating numbers against a quarter that already closed, and getting sign-off from a business owner who has no upside and all the downside.
So the migration handles the easy 60% and stalls on the rest. The old platform stays lit for the tail. The license renews. The savings that justified the whole program never land because the thing you were going to switch off is still running.
02Users follow familiarity, not roadmaps
The plan measures progress by what's been migrated. Users measure it by what they can find. Those are different numbers, and the gap between them is where adoption goes to die.
People go back to the tool they know, the report they bookmarked, the colleague who always has the right file. A migration doesn't change that reflex. It just adds a second place to look while the first one still works. You end up running both platforms in parallel, paying for both, and watching adoption of the new one stall because the old one still holds what people actually reach for.
Progress on the plan and progress for the user are two different numbers. The gap between them is where adoption dies.
03The platform that never dies
There's always one. Finance built its close process on Cognos a decade ago and will not touch it mid-year. A division came in through an acquisition running Tableau and its whole team thinks in Tableau. A regulated workload can't move without a compliance review nobody has budget to run this cycle.
Each of these is individually reasonable. Together they mean the estate never collapses to one tool, no matter how many planning cycles insist it will. 67%+ of enterprises run more than one BI platform (Datalogz, State of BI Report, 2025), and acquisitions and team preferences keep re-diversifying the stack faster than any single program converges it.
04What the sunk cost hides
The real cost of a stalled standardization program isn't the migration spend. It's the years of running a fractured estate while pretending the fracture is temporary. Users search five places and trust none. New hires learn the org chart before they learn where the numbers live. Nobody invests in making the current state good, because the current state is officially on its way out, forever.
That last one is the expensive part. You defer every access and trust improvement because it feels like polishing something you're about to throw away. Meanwhile the throwing-away keeps not happening.
05Fix access, stop waiting on convergence
Separate the user experience from the tool count. Put one front door over whatever you're running today: one search across every platform, certified content ranked first, ownership and recency visible before anyone clicks, usage measured across the whole estate instead of five admin consoles. Reports stay where they live and open in their native tool with native permissions. Nothing is migrated. Nothing is rebuilt.
Do that and two things change. Users stop paying the daily tax whether or not the migration ever finishes. And when you do decide to retire a platform, the usage data tells you what's safe to cut, and the retirement is invisible because people were never reaching that platform directly anyway.
Standardization can still be a goal. It just shouldn't be the thing your users' daily experience is waiting on. Give them the front door now, and let convergence run on its own timeline, decided by usage evidence rather than a date on a slide.
Show us your real stack, every tool, mid-migration and all. We'll demo against that, not a tidy sample.