Representative case study · TailorOS

How a Dhaka tailor shop digitised customer measurements in two weeks

The shop is in Mirpur 10, second floor of a small market block, two sewing machines and a fitting curtain. The senior tailor — we'll call him Karim Bhai — has been running it for twelve years. His assistant, Rakib, joined four years ago. Between them they do somewhere around 60 stitching orders a month, plus quick alterations.

The measurement notebook is a stack of three hardback registers, each filled front to back with customer names, phone numbers, and measurements in pencil. Karim Bhai can find any regular customer in under 30 seconds. He can also find them with his eyes closed — the muscle memory of where a customer's page is, top half or bottom half, register 2 or register 3, is real.

Why move at all

Three forces converged in early 2026:

• Two regular customers in three months asked for their measurements "by WhatsApp", because they'd moved to other neighbourhoods and wanted to order online from there.
• Register 3 was reaching its last 15 pages.
• A flood in March 2024 destroyed Register 1 (the oldest customers, 2012–2018). About 80 names were lost. Some were recovered from memory; most weren't.

Karim Bhai had been quietly resigned to "one day a flood will take everything". The 2024 flood proved it. When he heard about a shop two blocks away switching to an Android app, he asked.

What the shop tried first (and abandoned)

Before TailorOS, Rakib (the younger of the two) downloaded two generic "business CRM" apps from Play Store. Both lasted under a week.

The reasons were specific:

1. Account creation up front. Both apps wanted an email and OTP before showing the home screen. Karim Bhai doesn't use email; Rakib has one but doesn't remember the password. They never got past the signup.
2. No "measurement" field type. A tailor needs to record chest, waist, hip, length, sleeve, shoulder, and a dozen specific cuts. CRMs want to record "deal stage". The mismatch was instant.
3. Required internet. The wifi in the building drops 3-4 times a day during peak hours. Both apps showed loading spinners during those drops. Karim Bhai cannot say "please come back in 10 minutes" to a customer.

The TailorOS rollout, day by day

Day 1 — install, look around, nothing else

Rakib downloaded the app on his personal phone (an Infinix, 4 GB RAM). No account creation. The first screen asked for the shop name and the owner's name. That was it.

The first day was deliberately "no work". Just opening the app, tapping around, looking at the empty customer list. The trust-building cost of forcing a senior tailor to learn under deadline pressure is higher than any time saved by rushing.

Day 2 — first customer, written down twice

A regular came in for two new shirts. Rakib measured the customer (Karim Bhai watching), wrote the measurements in the register, then walked to the phone and entered them in TailorOS. The duplication was the whole point — the register stayed authoritative; the app was practice.

Then Rakib showed Karim Bhai the customer's page on the phone. Karim Bhai's first instinct was that the chest measurement looked wrong. They checked: it was right. The pencil in the register was smudged and looked like a 4 instead of a 41. That was the moment Karim Bhai got it.

Days 3–7 — every new customer goes in the app first

From day three onwards, every new customer's measurements went into the app first, then transcribed to the register at the end of the day. Old customers kept being looked up in the register; their data was not migrated.

By day 5, Rakib had added 14 customers. By day 7, the running list felt populated enough that searching the app was actually faster than flipping pages — and that's when the team started looking up old customers in the app first, even if they weren't there yet.

Days 8–10 — migrating Register 3 (the active customers)

Karim Bhai read names and measurements out loud from Register 3 (the current one), Rakib typed. They did this in 90-minute blocks between morning and afternoon rushes. Register 3 had about 240 active entries. It took just under three days.

Registers 1 and 2 — the historical customers — were not migrated. The judgement was that anyone who hadn't ordered in 18 months would re-give their measurements when they came back (because bodies change). This decision saved roughly two weeks of typing.

Days 11–14 — the cutover

By day 11 the active customer list was complete in the app. From day 11, the register was put away. The first three days without the register were anxious; everyone reached for it twice an hour by habit.

On day 14, they did the first "lookup test" — pick a customer at random from memory, find them in the app, read measurements aloud, check against the register. 8 out of 10 matched exactly. The other 2 were re-measured (the customer had been gaining weight; the app was now more right than the register).

What changed afterwards

One month after the switch:

• Customer lookup time: 30 seconds (Karim Bhai's muscle memory) → 8 seconds (typing first three letters of name in the app). Rakib is the same speed; Karim Bhai is faster on the app now too, which surprised him.
• WhatsApp sharing of measurements: zero a month → 6 a month. Customers who'd moved away started ordering again. Two of them paid in advance via bKash for the first time.
• Lost data risk: catastrophic (one flood) → low (encrypted backup to Google Drive runs nightly). Karim Bhai checked the restore process on his old phone after week three. It worked.
• New customers added per month: ~12 (pre-app) → ~18. Some of this is seasonal, some is the "the shop has an app" word-of-mouth among younger customers.

"I thought I would lose the speed I had with the register. I lost the smudges instead."
— composite reflection from Karim Bhai

What we'd do differently

Looking back at this rollout, two things would change next time:

1. Push the photo feature earlier

TailorOS lets you attach photos to a customer record — a snap of the fabric they brought in, or a final fit. This shop didn't use photos until week six. In hindsight, demonstrating the photo feature on day two would have moved the "wow" moment earlier and probably shortened the trust-building phase by a few days.

2. Migrate Register 2 selectively, not skip it

Register 2 (customers from 2018–2022) was skipped entirely. Over the next three months, four customers from that era walked back in. Two re-gave measurements without complaint; two were mildly annoyed that "you don't have my old size anymore". A 30-minute migration of the most-recent 50 entries of Register 2 would have prevented this.

The transferable lessons

This shop's specific situation isn't going to be your shop's. But three things generalise to almost every small-business digital migration:

1. Don't migrate history; cut over from today. Past data is rarely worth what it costs to type. The clean break is faster and the team trusts the new system sooner.
2. Build a daily ritual that proves the new system is right. In this case, the ritual was "look up a customer, read it out loud, compare to the register". Without that ritual, doubt accumulates silently.
3. Pick a tool that works during the worst moment of the day. For this shop the worst moment was the wifi dropping at 4pm. For another shop it might be the power going out. The tool has to work at that moment, or it doesn't really work at all.

The Bangladesh A2i programme has good public-sector data on digital adoption in small enterprises, and most of their failure-mode research lines up with what we saw at this shop. The reasons businesses bounce off digital tools are remarkably consistent: account creation friction, no offline mode, no recognisable domain fit.

TailorOS was built specifically to remove those three frictions. The shop above is one of many doing the switch — slowly, on their terms, without losing what already worked.