Every migration plan I’ve ever seen treats go-live as the summit. The Gantt chart crescendos to cutover weekend, and the boxes after it say things like “stabilization” and “hypercare,” thin little boxes, drawn by optimists. Then you go live, and you discover what those boxes contain.
After our big-bang platform migration at the major national retailer, described in the previous article, the system came up and couldn’t carry the load. An emergency fix bought us headroom, and what the headroom purchased was a state I call barely tolerable: the platform struggling, the business enduring, everything functional and nothing good. We lived there for weeks while the real repair, a months-long query optimization effort, ground forward.
What does the barely tolerable phase demand?
Operating a barely-tolerable system is its own discipline, and I’ve never read a methodology that describes what it’s like. I suspect the people who draw those thin hypercare boxes have never had to sit inside one. Three things defined it for us.
Triage becomes the operating system. When the platform can’t do everything acceptably, someone must decide what it’ll do badly.
Which workloads get priority, which reports can run overnight instead of on demand, which departments absorb the worst of the slowness. Those are business decisions wearing technical clothes, and they need to be made deliberately, by people with the authority to disappoint someone, instead of emerging from whoever complains loudest.
Letting the loudest complaints set the priorities is a leadership failure, and I’ve got no respect for executives who hide behind it. Somebody has to say out loud which department gets the slow screens this week, and then take the anger that comes with saying it.
Expectations need active management, daily. A workforce that just lost its familiar system and received a slow one in exchange will conclude the migration failed unless someone credible keeps telling them the truth: here’s why it’s slow, here’s what’s being done, here’s the real timeline. Silence gets filled with the worst interpretation, and morale-driven failure is real failure.
And you protect the fix from the firefight. The same people best qualified to build the permanent repair are the ones everyone wants on today’s emergency. We had to deliberately wall off the optimization work from the daily triage, because a team that spends every day surviving never ships the thing that ends the survival phase.
People also keep a ledger in those weeks that appears on no dashboard: credibility. Every honest daily update deposited a little; every workaround that held deposited more. The reserves mattered because barely-tolerable operation runs on discretionary human effort, the accounting clerk accepting the slow report, the manager not escalating, and people extend that effort to teams they believe.
A degraded system that everyone just accepts taxes every employee who touches it for years, and nobody ever adds up the bill.
We could spend trust because we had been earning it daily. I’ve watched other organizations run the same weeks with an empty account. Every slowdown became an escalation and the recovery drowned in its own meetings.
The trap in the phase
The subtle danger of barely tolerable is that organizations acclimate to it. Week three, the workarounds have hardened into routine, the complaints have gone quiet, and the pressure that funds the real fix begins to leak away. Barely tolerable starts to look like the new normal, and I’ve watched companies, not mine, thankfully, stall there permanently, running degraded systems for years because the crisis lost its urgency before it lost its cause.
That outcome is worse than a failed migration would be. A failed migration gets rolled back and argued about. A degraded system that everyone just accepts taxes every employee who touches it for years, and nobody ever adds up the bill.
The countermeasure is measurement. We knew what healthy looked like in numbers, we knew where we were against those numbers, and the gap stayed visible to the people writing the checks until it closed. Barely tolerable must be a bridge with a far end, and someone has to keep pointing at the far end.
My take
Treat the weeks after go-live as the real project. Decide out loud which department gets the slow screens this week, and take the anger for it. Tell the workforce the truth every day, because silence gets filled with the worst story. Wall the people building the permanent fix off from the daily firefight. Measure the gap to healthy and keep it in front of whoever writes the checks. Barely tolerable has to be a bridge, and somebody has to keep pointing at the far end.
Go-live is the midpoint of a migration, and nobody plans for the second half.Share on X
Why does the go-live story belong in your book?
Executives who write about their transformations reliably skip this phase, straight from the dramatic cutover to the happy retrospective.
That omission is a mistake, because readers who’ve lived a migration know exactly what got skipped, and they trust the author less for it. The barely-tolerable weeks are where the leadership happened: the triage calls, the credibility spent, the fix protected from the fire. I’d tell any executive planning that book to give those weeks a full chapter, because practitioners will judge every other page by whether that one rings true.
For more from this series, see The Digital Transformation Hub: real transformations, lived from the inside, decades before the term existed.
