Skip to the case study

All case studies

Jill Low

Product Designer, UX/UI & Front-end Coder

Case study · Customer identity Senior UI/UX Designer, StoreHub Nov 2018 to Dec 2020

Merging four customer databases into one identity

StoreHub's point of sale, online stores, and three new Beep products each kept their own version of a customer, so the same person could show up as four different records with no link between them. We merged the online store's customers into Beep's, using a mobile number and a one-time password as the identity everything else pointed back to.

01

Four products, four kinds of customer

StoreHub started as a point of sale system. Merchants added their customers straight into the point of sale or the BackOffice behind it, the same place they managed products, inventory and reporting.

We then added e-commerce: merchants could instantly publish an online store from the products already in their POS, and their customers could buy from it either logged in or as a guest.

Within the past year, StoreHub had also launched three consumer products: Beep Delivery for online food orders, Beep QR Order for table ordering, and Beep Cashback, a loyalty programme that gave customers a percentage back on each purchase to spend later. Beep identified a customer by their mobile number and a one-time password, rather than a name or account they set up themselves.

02

Four records, no link between them

Every new touchpoint had added its own version of the same problem:

  1. Point of sale customers could be freely added by anyone at the POS, so the data behind them wasn't reliable.
  2. Beep Delivery, QR Order and Cashback captured no name or email at all, only a mobile number, so merchants knew even less about these customers.
  3. If every channel's customer data were pulled together in the BackOffice, the same person would appear multiple times, inflating a merchant's real customer count.
  4. Beep customers were already linked to point of sale customers, so one Beep customer could correctly link to several merchants. No such linkage existed for online store customers at all.
Diagram showing Merchant A and Merchant B's point-of-sale customers each linked to a shared pool of Beep customers, while a separate, unconnected pool of online store customers sits apart with no link to either.
How the four kinds of customer data related to each other. Online store customers were the one group with no link to anything else.
03

Merging into Beep's identity, not building a fifth one

After discussions with the product managers and our CTO, the simplest fix was also the one that avoided a fifth customer model: retire "online store customers" as their own segment and extend Beep's database to support them instead. That meant a few specific rules:

  1. Existing online store customers could still log in with email and password or Facebook, but every new customer from here on had to sign up with a mobile number and OTP.
  2. Existing customers were encouraged to add and verify a mobile number after checkout. Once verified, they could log in either way.
  3. A verified mobile number couldn't be changed on an account afterward.
  4. To claim cashback in person, a customer gave the cashier their mobile number.
  5. A customer shopping from an online store could also claim cashback, if the merchant had Beep Cashback turned on.
  6. We planned to eventually remove guest checkout on online stores, so merchants would have real data on every customer who bought from them.
Flowchart of the online store checkout: a customer can log in or sign up with a mobile number and OTP, log in with email, authenticate with Facebook, or check out as a guest, before selecting delivery, paying, and confirming the order.
The checkout flow after the merge: four ways in, one identity underneath.
04

What it looked like

After a lot of iteration and back and forth with the team, this is what the login and checkout screens ended up looking like.

Video: the online store's login screen, offering email and password, mobile number, or Facebook to sign in, with a link to check out as a guest below.
The finished login screen, with the cart summary carried alongside it so checking out stayed one continuous flow.
05

More customers chose to log in

Because this touched every merchant's customers at once, the product manager watched it closely after launch in case it needed to be rolled back. It didn't.

Bar chart of logged-in versus guest checkouts on online stores from November 2018 to August 2020, with logged-in checkouts becoming the majority from June 2020 onward.
Logged-in versus guest checkouts by month. Logged-in checkouts overtook guest checkouts after the June 2020 launch.

The product manager also ran an A/B test on removing guest checkout entirely, which was still running when I last had visibility into it, so I can't say how that landed.

Two things were still in progress by then: letting customers spend their Beep Cashback balance directly on an online purchase, and dealing with the customers who'd end up with the same mobile number attached to more than one email. The plan for that second one was to let merchants merge those records automatically, favouring whichever touchpoint held the most reliable data.

Video: a mocked-up customer record showing the proposed merge behaviour, combining a customer's point of sale and Beep history under one profile.
The proposed merge behaviour, mocked up with test data — not yet built when I left.