Mintu bhai used to drive a Pathao bike before he started moving apartments full time for a living. He says the bike was simpler, two wheels and one helmet and one passenger who mostly wants you to stop talking and get to Gulshan-2 before the meeting at 11. We'd been waiting for him at the lift on the fourth floor in Banani 11 since two o'clock that Friday, the ammu fanning herself with a folded BRTC schedule and the abbu pretending the heavy mirror in the corner was lighter than it was. Mintu bhai showed up at 2:14 with a roll of brown packing tape on his wrist like a bracelet, a fat Camel marker behind his ear, and the calm of a man who has done this seventy times.
The truck part took ninety minutes. The labelling took the rest of the afternoon. 47 cartons came out of that flat by 4:15pm, and not one of them landed in the wrong room at the new place in Banani 5. What Mintu bhai actually got paid for was knowing that the ammu wanted her kitchen unpacked before the bedroom, that the abbu would not admit until day three that he could no longer lift the heavy mirror, and that the kids' school books had to be findable on Monday morning when the alarm went off at 6:40 for Sunbeams.
That, more or less, is the silent half of product management. Most product writing is about the v2 launch, the redesign, the new pricing page in Helvetica. The actual unglamorous skill is moving fifty thousand existing users off v1 onto v2 without dropping any of them off the back of the truck on Bashundhara Road. Marty Cagan has been writing about this in Inspired under the heading of transformations, and he's right that it's a different muscle from feature work. Lenny Rachitsky's newsletter had a sharp piece on rollout strategy whose headline finding is that teams overpolish the launch email and forget to write the deprecation email at all. Patrick McKenzie made the customer-success version of the same point in a post on enterprise transitions, where the bank that handled the cutover with grace got six more years of contract and the one that didn't lost the renewal in a quiet email on a Tuesday.
What all three of them are circling, in their different vocabularies, is that a migration is a logistics problem with a psychology problem inside it, the way a Friday afternoon in Banani is a truck problem with a family problem inside it.
In my own count there are roughly five moves in any migration worth the name, and I'll walk through them the way Mintu bhai walked through the boxes that Friday. None of the five are exotic, and most v1 to v2 launches I've watched go badly in the last six years went badly because at least one of the five was treated as somebody else's job.
all forty-seven cartons in
Last carton in. Mintu bhai checks the manifest one more time before he closes the cargo door.
The order of those five matters more than the labels do. The default has to be settled before anything downstream can be sequenced, because every other decision is conditioned on what the unattended user will end up with. Opt-in cadence comes last because you can't slot cohorts onto a calendar until you know which deprecation grief you're managing each cohort through.
The bit that gets PMs into trouble most often, in my own honest count, is dark launches. Most teams treat the feature flag as an engineering concern and the rollout curve as a marketing concern, when the rollout curve is a product artifact that belongs in the same doc as the spec. A flag at 1% for 48 hours, 5% for a week, 25% on a Wednesday afternoon when most of the eng team is in the office, 100% three working days later if the dashboards held. The curve has opinions in it about who you trust to find what early, and those opinions are part of the product.
There's a Bangla phrase Mintu bhai used twice on the lift landing, gucchiye nao, which roughly translates to "tidy it up properly before you carry it." It came up the first time about a tray of tea about to be wrapped without being wrapped, and the second time about the v2 onboarding flow I'd been describing to him between floors. A v2 launch that hasn't been gucchiye-nao-ed at the migration level ships as a feature with a thousand quiet defects shaped like users who never quite finished moving in.
What the boring habit gives you, across three or four migrations, is hard to describe without sounding pious. The team starts thinking of the user base as a thing with weight and shape, the way Mintu bhai thinks of 47 cartons as a manifest with a sequence and a destination. The retro question stops being whether the v2 shipped on time, and starts being whether all 47 cartons landed in the right room by Monday morning when the kids needed their books.
A migration that ships well looks suspiciously like nothing happened, and that is a career problem before it is a craft problem. The PM who runs three of them in a row comes up for promotion on a deck that says "drove key transformation initiatives," which is both true and useless, and which undersells the work in precisely the way the work invites.
Mintu bhai's invoice for the day in Banani 11 said "labour, transport, packing materials." It did not mention the ninety minutes he spent labelling cartons on the lift landing on the fourth floor while the ammu's hand fan kept going and the abbu kept pretending about the mirror. He did not think to charge for those ninety minutes, and if I had suggested it he would have been embarrassed, because in his trade the labelling is not a service, it is just what you do if you are any good. I have never worked out whether that is admirable or whether it is the reason he still quotes by the truckload.