adidas Member Login

Challenge

Adidas had identified membership as a major opportunity to create more value for both customers and the business. Working with Product and Data, we uncovered a particularly interesting behaviour in checkout: existing members who successfully logged in converted significantly better, yet many returning members were either ignoring login or failing to complete it.

Approach

I led the design work to understand what was getting in the way and how we might change that behaviour without adding friction to checkout. Working closely with Product, Data and Engineering, we explored several hypotheses, from passwordless login to a more personalised experience and used research and experimentation to understand which ideas actually improved the journey. Not everything worked, and one seemingly promising solution taught us something particularly important about what “frictionless” really meant for our customers.

Outcome

Ultimately, the simplest intervention proved the most effective. Showing customers only the login method they’d previously used increased Login Success by 19.63% and logged-in visits by 15.10%, with an estimated €1.05m in incremental annual net sales across EU Big 7 + US.

01

The opportunity

13% of delivery-page visits were detected members with a password.

Of those, 48% ignored or closed the login overlay. Of the 52% who attempted login, only 54% succeeded.

And successful logins converted roughly 20% better than those who tried and failed, and 10% better than those who ignored the login overlay.

1. 13% of delivery-page visitswere detected members with a password

2. 52% tried to log in48% ignored or closed the overlay

3. 54% of those succeeded46% tried and failed

Best-converting cohortMembers who logged in converted ~20% better than failed attempts, and ~10% better than those who ignored the overlay.

There was clearly value in getting more existing members logged in. The question was how.

02

How we framed it

We didn’t start by designing a login component.

I worked with Data Analysts to map the journeys and identify the opportunity.

Then I worked with the PM to design a workshop around:

• framing the problem with evidence
• understanding the business opportunity
• bringing Engineering into the problem early
• generating multiple solutions
• identifying unknowns and risks
• defining what success should look like

I asked our Experience Director to also participate and present an overview of the ‘Lower funnel’ strategy so the developers would understand our goals and what we want to achieve from a higher level. It worked well and created a sense of engagement and ownership from the team which was felt later on in the tougher stages of development.

In this workshop, I wasn’t simply asking developers to brainstorm UI. I tried to give them the lower-funnel strategy and business context first, so they understood what they were being asked to help solve.

One of the things I deliberately did during the workshop was document the unknowns alongside the ideas.

Rather than treating the workshop as the end of discovery, we left with a list of questions to answer.

For example:

• Why couldn’t 46% of users log in?
• What did we already have available for social login?
• What could we leverage from existing Magic Link functionality?
• How sensitive were customers to using their phone number?
• What were the business risks of choosing the wrong intervention?
• What would success actually look like?

These became concrete follow-up actions for the wider team.

This helped us avoid designing around assumptions simply because they had emerged from an enthusiastic workshop.

Rather than choosing the idea that felt best, we captured the unknowns and designed experiments to reduce them.

03

Three hypotheses

Ultimately, the simplest intervention proved the most effective. Showing customers only the login method they’d previously used increased Login Success by 19.63% and logged-in visits by 15.10%, with an estimated €1.05m in incremental annual net sales across EU Big 7 + US.

04

The first twist: the exciting solution

HypothesisPassword friction is preventing members from logging in. If we remove the password, more people should successfully log in without damaging checkout conversion.

The experiment ran for 57 days in the US with full members on the Delivery page:

The experiment showed that Magic Link could increase login success, but users had to leave the checkout journey to retrieve the email. Only 48.62% of those attempting Magic Link successfully logged in.

And 7.15% clicked “Email me Magic Link”, meaning the feature wasn’t being used by a large enough share of users to compensate for the friction it introduced.

We solved the login problem… and created a checkout problem.

05

The decision

The team recommended not rolling out Magic Link, despite its positive effect on the primary login KPI.

The experiment changed our definition of success. Login Success wasn’t enough. We needed to improve login without compromising the journey around it.

The evaluation estimated that testing the feature prevented roughly €1.03m in potential revenue loss.

So even though the test sort of ‘failed’, it generated valuable evidence and protected the business from rolling out a quite harmful change.

06

The second experiment: make login personal

We moved from asking“How can we invent a better way to log in?” → “How can we make the existing way feel easier?”

The new hypothesis was to make the login component more personal by only showing the previously used login method:

1. Experiment 1Magic LinkRemove the password
2. vs.
3. Experiment 2Previously used methodRemove the decision

07

…And this one worked!

The experiment also showed the email → login → order funnel improving by +21.45% within one pageview to order.

And importantly, most experience KPIs remained flat, while Payment → Exit showed a directional −2.56% change.

So we didn’t just improve the headline KPI, we did so without introducing the downstream checkout damage we’d seen with Magic Link.

In summary:

08

Learnings

Local optimisation can make the overall journey worse

Magic Link increased login success, but negatively affected downstream checkout behaviour.

The experiment exposed the risk of optimising one part of a funnel in isolation. From that point, we broadened our definition of success beyond authentication and evaluated solutions against their impact on the whole checkout journey.

It also challenged our original assumption. The problem wasn’t necessarily that passwords created too much friction.

Customers may simply have needed more confidence about which login method to use.

Bringing engineering into discovery early exposed capabilities we hadn’t initially considered, including existing authentication technology we could use to test Magic Link quickly. This allowed feasibility to shape our hypotheses before we invested heavily in design.