Postpaid to Prepaid Migration
500,000 customers, and none of them could stay where they were.
01 / THE NUMBER
Five Hundred Thousand
That was how many StarHub customers had to move from postpaid to prepaid. Not could. Had to. The new plan would not run on the old app, so there was no version of this where a customer opened the app they knew and carried on as before.
The business framed it as a free upgrade. 5G at no extra cost, better roaming, better IDD. All true.
Customers do not experience plan changes as upgrades. They experience them as something being taken away, dressed up. Prepaid in particular carries a downgrade connotation, and telling someone they are getting more while moving them somewhere they did not ask to go is how you manufacture suspicion at scale.
Half a million people. If even a small fraction of them arrived confused, that becomes complaint spikes, call center load, retail queues, and churn. The cost of a bad migration is not measured in design. It is measured in headcount.
02 / WATCHING, NOT ASKING
Not a plan change. A system change.
The gap between how the business described the task and what it actually required is where migrations go wrong.
The business description: your plan is being upgraded.
What the customer actually had to do: understand that something was changing, receive an SMS, find and download a different app, log in through a new identity flow with mobile number and OTP, and then verify for themselves that their plan, their usage, and their credit balance had all survived the move.
That is not a plan change. That is a change of system, identity flow, and mental model, and every one of those steps is a place where a person can stop, get confused, and pick up the phone.
03 / WATCHING PEOPLE, NOT ASKING THEM
We ran research where the customers actually are.
I worked with my researcher to plan and run moderated sessions on the ground, in prepaid reseller environments rather than a lab. Prepaid customers are not a monolith, but the segment skews toward lower language literacy and lower technical confidence than the postpaid base, and you do not learn how that plays out by putting people in a usability suite and asking them to think aloud.
Two findings reshaped the work.
People do not read. Text-heavy instructions were skimmed or ignored outright. Users with lower language literacy closed pop-ups fast and tried things instead, navigating by trial and error rather than by reading. Whatever we wrote, most of them were not going to read it. Which meant clarity could not be solved with copy. It had to be solved structurally, with visual hierarchy and sequence.
People misread their own dashboard. Wallet balance, data balance, credits: users confused them, and mistook them for the main state of the account. Payment and top-up flows made sense only when they appeared at the moment of need. Nobody proactively sets up a payment method. They set one up when they are about to run out.
Both findings pointed the same direction: progressive disclosure, contextual prompts, guidance sequenced to behavior. Not more information. Less, at better moments.
04 / THE ROLLOUT WAS PART OF THE DESIGN
10K, then 190K, then 300K.
We staged the migration in three waves across three weeks. Ten thousand customers first, to test whether the systems and the journey held. Then 190,000, once we had learned something. Then the remaining 300,000.
This is usually filed under delivery, not design, but it is a design decision. A staged rollout is what lets you find out that your journey is wrong while it is still cheap to fix. Trust gets built by reducing uncertainty, and that applies to the organization shipping the change as much as to the customer receiving it.
05 / WHAT WE BUILT
Meet people before they are confused, then confirm fast.
I designed the core migration flows myself, working with UX and visual designers on the wider launch plan.
A heads-up inside the old app. An in-app modal announcing the 5G upgrade, framed explicitly as free, with a single path forward: find out more. The framing was deliberate. Suspicion is the default response to a change you did not request, and "free" is the word that gets a customer to keep reading rather than start complaining.
An information hub instead of a wall of text. The modal led to a dedicated page: what the upgrade gives you, then an explicit sequence of what happens next. Expect an SMS. Download the new app. Log in with your number and an OTP. Check that your details are right. Plus an FAQ covering the four things people were actually anxious about: their credit balance, their SIM, how to top up, and who to call.
FAQs on a program this size are not filler content. They are load-bearing. Every question answered on the page is a call that does not reach the contact center.
Onboarding that knew who you were. The new app could detect a migrated user at login and trigger a tailored flow for that cohort. Tooltips pointed them at one thing: go into the Mobile section and see your plan, your usage, your credits, intact.
This came straight out of research. Reassurance does not come from reading an explanation. It comes from seeing your own balance still sitting there. So we optimized for time-to-confidence, not comprehension. Get them to the proof as fast as possible.
06 / WHERE IT LANDED
It rolled out after I left.
The program went live after I had gone. I do not have the post-launch numbers, and I am not going to imply otherwise.
What I can say is what shipped: a complete journey across four touchpoints, staged for risk, with the messaging, the information architecture, and the first-login experience all traceable to what we watched customers actually do in a reseller shop.
07 / REFLECTION
What I'd do differently.
I had the authority to push for a real awareness campaign, and I didn't use it.
The heads-up in the old postpaid app was the thinnest part of the whole program. It should have been the thickest. That app was the one place we knew every affected customer would already be, and we had months before the first wave to use it. Instead we treated awareness as a modal, and then a page, and then we moved on to the part everyone was watching.
The reason is not that I was overruled. It is that awareness work does not look like the job. The migration was the deliverable, with three hard waves and a fixed date, and building a slow campaign inside an app we were about to retire felt like scope creep. Nobody thanks you for the complaints that did not happen. So the work with the least visible payoff got the least attention, and I let it.
That was my call to make and I made it wrong.
The value proposition was never exciting enough to carry a forced move.
"5G at no extra cost" is a fact. It is not a reason to feel good about being moved somewhere you did not ask to go. We had a genuine opportunity to make this feel like something customers were getting rather than something happening to them, and we settled for accurate rather than persuasive. Research told us these customers were suspicious by default. Accuracy does not beat suspicion. Desire does.
The real lesson: the migration begins before the migration.
By the time our first modal appeared, most customers had already decided how to feel about this. We designed a very careful arrival, and under-designed the departure. If I ran this again, the sequence would run in reverse: earn the anticipation first, in the app they already trust, over weeks rather than a screen. The transition itself is the easy half.
