Increasing new memberships with rapid experimentation

This is the story of how I supported a new disruptive startup in the beauty sector to gain an uplift of 44% for new membership conversion. This was done with qualitative user research and a lot of persistent iterative prototyping.

The challenge

This one-year-old disruptive start-up in the beauty sector had experienced amazing growth.  They had a membership club model that gave members access to non-branded lower cost, high-quality beauty products.  They’d had an early surge in new memberships on launch and were now looking to grow this to a much wider market.  While they had an initial uplift of new users, these were mostly early adopters.  The brand was now struggling to convert to a wider audience of newer customers.

THE IMPACT

44% uplift for new membership conversion

Since my working with BeautyPie as a a direct result of the customer journey I designed, there was an uplift of 44% in new membership conversions.

New customer journey and updated checkout

The Beauty Pie brand now has a clear the customer journey for new and existing members.

New brand terminology and styleguide

A tested brand terminology style guide to ensure consistency of messaging and content across channels.

As with many startups, they had run fast to kick off quickly but didn’t have the data available for meaningful site traffic analysis.  They had also been burnt with expensive consultants and sluggish web development partners who couldn’t get to the root cause of the issue.

The team had lots of internal speculation for the reason of the lower membership sign-up.  Mostly this was seen as an awareness problem and that they just needed to drive higher volumes of people to the site. I was going to need a way to ‘get out of the building‘ and really understand what was happening for customers. 

Watching customers use the website 

The simplest, fastest way I know how to understand what is happening is to watch customers using the website.  I set about immediately recruiting new participants for this study.  I needed access to a pool of people that hadn’t heard of the brand yet.  This is pretty straightforward if you use the services of a people finder market research agency.  

To learn why the website wasn’t performing, I needed to observe participants over the whole journey.  What were their expectations before they arrived on the site, why did they come to the site initially?  What was their goal, and what was the process they went through in trying to achieve it?

We created a scenario that tried to emulate the beginning of the journey that most people might take.  The people selected for the test had already indicated an interest in beauty products as something they do with their time.  We either gave them an article or social media promo post that talked about the startup for the user test.  I asked for their immediate reactions and understanding.  This helped me build a picture of their motivations, messaging that drew their attention, and how much information they could retain.  In every case, they indicated they liked the proposition of the business and would want to learn more.  I invited them to follow this interest with whatever means they had available.  Most of them whipped out their phones and either jumped on google to learn more or went direct to the brand’s website.

Crucial insights from early user testing

In observing 11 people (1 no show) in my test scenario, only 1 out of 11 participants actually understood how the proposition worked and wanted to sign up.  I edited the live recordings of the users on the site into a hard-to-watch highlight video. For the team, this was difficult for them to see.  It was the first time they had been confronted with the reality of how their hard work was being received outside their loyal customer base.  

However, it did give us a starting point as to why customers may not be signing up.  With each user test, I started to see repetition from the same emotional rollercoaster.  I created a visual map of this emotional journey below.

This visual map of the emotional journey was a helpful tool for creating a shared understanding of how new customers might be experiencing the brand.  It showed that while the proposition of the brand was strong, customers were not understanding it.  This emotional crash of misunderstanding after investing time exploring was the killer frustration point.

The new customer identified behaviours.

I continued to see patterns and groupings emerging through user tests.  New customers seemed to fall into two groups with a primary goal when determining if they would consider signing up to be a new member.   This was key to realising that we would need to create journeys that cater for both sets of user needs.

As I observed more user tests, I could further identify more customer needs into a ranked order of importance.  While this was based on observations, it provided a base of assumptions that I can set about validating.  It seemed like understanding the proposition was the first hurdle to anyone wanting to sign up.  Then other needs followed.

Customer testing and iterative design

Having understood the challenges and identified that explaining the proposition was the most important customer need, I designed a one-page prototype.  I wanted to create a landing page that would immediately meet the customer need – “I want to understand how it works before I look at anything“.   I looked at other websites that explained their proposition on one page.   Using the language I’d heard people use, I sketched out some rough content that might explain the proposition.  It took about 1-2 days to refine and put onto a web page until it was good enough.   I bought in a new set of new customers and repeated the scenario, only this time I rigged the laptop and phone in the room to reroute to my prototype (without the users knowing). 

No one reads anything.

The first test revealed that no one reads anything unless it is very, very succinct.  Getting that balance right was the key.  They skimmed the first bit of content and closed the page.  It was hard to see my work get so quickly dismissed, but it only cost me a day or so to try. 

My next idea was that perhaps a video explainer would work.  I created a basic slideshow, exported it as a movie, and stuck it in on my next prototype.  The next batch of users came, who looked at the video for about 3-4 seconds and closed it.  This is the Instagram and tick-tock generation.  If a video doesn’t hold their attention in the first 2-3 seconds, they are gone. I could also see that the users weren’t yet invested in watching a long(ish) video at that early point in the journey.  They just wanted it all explained in 3-4 seconds.  I ditched the video and went back to just text and images but tried borrowing some of the simplified messaging I’d created for the video. 

At last, the participants were starting to get it.

Success! Most participants in this round of testing started to get it.  This was working well for the journey of customers that wanted to learn everything before exploring.  I now needed to focus on the other set of users that “don’t want to read anything, and just want to dive in and work it out as they go“.  

During the tests, I started to see that key terminology across the site was varied and inconsistent.  It was creating confusion and wasn’t perhaps the right choice of words. I asked users to sort and feedback on keywords that described the service.  This then informed a new content style guide for terminology that I had tested.  The style guide allowed the content team to start harmonising new output immediately.  These changes, followed by some core product page and interface tweaks, started to nudge the rounds of participants over the line in ever more consistent numbers. 

I tested 5 different prototypes with over 50 new customers. After each test I took the learnings and insights from each test to iterate and develop the designs. This whole process took about 8 weeks with no development and about 40 customers.  

The team could then move forward with expensive web development with higher degree of confidence that the design changes would have an impact.  While it’s impossible to ever know for sure we had massively de-risked making any major blunders, this gave the founder enough confidence to make the changes.  The impact was fairly immediate with an uplift of 44% in new membership conversion.

You can read more about this story from the founder.