In a nutshell

This story about designing registration flow contradicts two famous dogmas:  'If it works, don't touch it' and 'Less is more'.

We didn't touch a single thing and observed our conversion rate decrease by 30%; counter-intuitively, we resolved that issue by adding an extra step to the registration process.

Role: Design & Product Lead
Timeline: Form improvement — 1.5 months, total project timeline — 2 years
Teams & parties involved: Campaign Management, Design, Engineering, Legal, CTO, PMs, Content
Context & Goals
Understanding the problem
Finding solution
Launch
Fixing the issues
Outcomes

Context

On December 13, 2017, Alexey Navalny, opposition leader in the 2018 Russian presidential elections, announced the start of his campaign.

I led the campaign website, product, and branding teams through a 2-year electoral campaign. And, oh boy, do I have a lot to share. But let me focus on one little story here — the registration flow (aka onboarding).

Goals

Our main objective was to register people who would officially support Navalny's nomination. We needed to register at least 300,000 supporters.

1
Register 300,000 supporters
2
Make sure those registrations were from real people

Step 1. Understanding the problem

Electoral context

Every independent presidential candidate who wants to get on the ballot needs to collect 300,000 signatures from Russian citizens. These signatures must be collected and verified within 3 weeks, on paper, and in at least 40 different regions (there is a limit of 7500 signatures per region).

From a legal and technical point of view, this process has a nightmare level of complexity. To navigate through this complexity, we had to register, locate, and verify our supporters in advance.

To simplify — it was crucial for us to know that we had enough supporters and to know where they live.

Specific problems to solve during registration

Step 2. Finding a solution

Information needed & potential issues

We tried to simplify the form as much as possible, despite the hefty amount of information we needed to get.

Optimizing the input fields

Click to zoom in

Form layout

Trust me, we experimented with different layouts before we settled on the final look. We didn't want to split the form into multiple steps or for it to look overwhelming when presented on one screen.

As you can see, even the simplified version of our form was tricky and had a lot of fields. Our next challenge was to make it very discoverable and at the same time not scare people away with a gigantic one.

Registration form desktop layout

Discoverability & accessibility

At the early stages of the campaign, registering supporters was the main aim of the website (and arguably of the campaign itself). With that in mind, we built the entire website — every single page had a very visible call to register right in the header.

To increase our chances of getting more contacts, we decided to split our giant form into two steps. At the first step, we would only ask for an email, and at the second step, the rest of the form would be shown. This way, even if it was not a good time for someone to fill in our entire form, we could still get in touch later.

✨DEVIOUS PLAN✨ (It worked.)

To make the form more accessible we used contrasting colors, thoughtful error messages, and optimized our UI for mobile, screen-readers, and keyboard navigation.

Registration flow

1
Lazy registration (email only)
2
Registration form + phone confirmation
3
Registration confirmed

Testing our solution

Because everything was done in a very secretive way, A/B tests were off the table. Before the official launch, only a very small group of people involved were aware of the project. However, we did a quick usability test on a small group of employees not engaged in the design and implementation process.

Step 3... 2, 1, Launch

Due to unforeseen circumstances and the political situation in Russia, we had to launch 6 weeks earlier than planned. This left us only a week to regroup. After some fast and furious adjustments, heated discussions, and 48 sleepless hours, the website was up and running just on time.

Everything seemed to be working smoothly, we survived the initial volume spike of registrations, hundreds of thousands of supporters were going through our form with seemingly no problems at all. All our conversion numbers looked nice and healthy.

What can go wrong when everything goes so well?

‘If it works, don’t touch it’ — I bet you’ve heard this ‘golden rule’ of software development before. Apparently it doesn’t apply to product development. Imagine our surprise when after ~2,5 weeks of ‘not touching it’ we noticed that our conversion rate from step 1 (enter email) to confirmed registration (form fully filled, phone number and email confirmed) dropped from ~51% to 34%.

Nothing had changed, we hadn’t touched a single thing, and yet somehow our key metric dropped significantly.

We started our investigation and quickly noticed that a significant number of supporters filled in the form, confirmed their phone number but failed to confirm their email. That seemed odd. Filling in a giant form, sharing personal info and verifying your phone via text message involves significantly more effort than one-click email confirmation. Why commit to the first part and then skip the easiest step?

We quickly discovered that our users didn’t see any value in email confirmation. On top of that,  we realised that we had managed to send out probably the lamest ‘confirm your email’ email of all time — it had a technical nonsense title, and didn’t explain the value or urgency of this action. From the very beginning, we were sending out a placeholder created for a technical test, and somehow everyone overlooked that.

Why did everything go so well if everything was so wrong?

It turned out that the real mystery of this project was the reason why we had our very decent numbers for the first two weeks. Somehow 59% of people managed to fill in the form and confirm their emails, despite our sins.

Our guess was that the magic was in the audience. Our first visitors were our hard-core supporters and long-term fans. Super excited and motivated, they followed instructions carefully and paid a lot of attention to all communication from the campaign headquarters (even the lame emails). But as time passed, and the campaign started to reach a broader, less involved, and less motivated audience our product failures became more obvious.

In case you are wondering how lame an email can be:

Step 4. Fixing the issues

We fixed the lousy email, but we couldn’t get to our initial metrics.

At the same time, our data showed that the majority of people confirming their emails did so within an hour of filling the form. The longer they waited, the less likely they were to finish the flow and confirm their email.

When more is more

It turned out that the email itself was not the best tool to convey the urgency of email verification. If a person was unfamiliar with the process or simply didn’t go and check their inbox promptly, our chances of successful confirmation decreased.

Since there were no other interaction points between filling the registration form and email confirmation, we decided to extend our registration process and add one more step that would nudge supporters towards email confirmation.

We added an extra screen clearly indicating the registration process was not finished. The page presented a short check-list with 2 out of 3 check-boxes checked and only one remaining — 'confirm your email'. We explained the importance of confirmation, and to make things even easier, we added a huge red button that would take you to your inbox straight away.

Updated registration flow

1
Lazy registration (email only)
2
Registration form + phone confirmation
3
Check list + CTA to check your email
4
Registration confirmed

Outcomes

  • With an initial KPI of 350,000, registered over 700,000 people ready to support the nomination in 6 months (and over a million within a year)
  • Conversion from lazy registration (email only) to fully confirmed account increased 51% 70%
  • Conversion from filled form to fully confirmed registration increased 70%89%
As a side note: we were able to convert about 10–15% of 'lost' users retrospectively with our follow-up sms + email campaign

Lessons learned