Helping 10% more customers through an account application form
By rebuilding the account Application form at Virgin Money, we significantly lowered the companies Cost per Acquisition for Savings products - the same form is still in place now, still saving money each year.
1. Exploration & research I knew something was wrong. I didn’t know HOW wrong.
Competitor analysis – We came out as bottom of the table amongst our peer banking group. Thier processes were shorter, simpler and a lot quicker to get through. Ours was officially confirmed as awful.
Analytics snapshot – I looked at our analytics and took a deep dive into what the numbers were telling us. Device, time on page, overall time to apply, conversion levels, steps in the process (up to 29!), revisits and abandonment. All pretty ugly.
Voice of the customer – Call listening, feedback, complaints, you name it, I got hold of it. Really really listening to the customers who would call us every day. Stuck. Having dashed their hopes and dreams of a simple, usable application form.
At this point, I was more of an undercover UXer. My official title was now ‘E-Commerce Manager’ so a lot of what I was doing was on a wing and a prayer. Without access to a research budget and with this in mind, I needed to work closely with another UX designer because at was at this point, I started to really understand the scale of the problem. And it was huge. Internal research – as we suspected, it wasn’t just customers who had an issue with our application form. Our colleagues did too. We uncovered some very surprising points.
Who knew that we were spending lots of money checking the identity of each customer who came through the application, whether they were an existing customer or not?
We spent a lot of time mapping it out and understanding which questions were needed to open an account.
We then held a collaborative design workshop with some key stakeholders to understand their ideas and come up with some early ideas of what needed to change. By also focussing on a mobile first methodology, no one could get caught up in the design.
I reviewed content and took out irrelevant questions and brought the process down to a dozen simple questions. I then looked at chunking the information into understandable sections and even glued some questions together.
I then looked at simplifying the words used - how many people really understand financial terms? This was great progress. Filled with ideas and enthusiasm, we began to sketch and re-write.
I took a little inspiration from other industries and worked out how to bring the registration process into the application, much like the idea in an e-commerce site of offering a simple registration after the point of purchase, and put it into a low-fi Axure wireframe prototype. This wasn’t taken forward with the project due to technical constraints, so I refocused to look at how to simplify just the application on it's own.
We checked our design with some users who weren’t familiar with the process. We refined, then took this back to stakeholders. Design, test, refine, repeat.
One of the many interesting anomalies which were uncovered in our discovery phase was this;
Question: Please provide an existing Virgin Money account number.
Simple enough you might think. We assumed (incorrectly, as I am sure users also assumed) that in the background, we would be making checks against this account number to match up the users information, confirm if they had already passed ID & V checks, and then help them through the rest of the form by filling in what the system already knows.
None of this was true.
Every time a user went through the application, whether they had indicated they were an existing customer or not (by signing in or giving us an existing account number), we sent off all of their details to Equifax, at the approximate (tiered) cost of £1 per check.
If we made a simple change so that even those users who were logged into their online system prior to beginning the application (roughly 27% of users), we could save around £15,000 per annum in third party ID and V checking. In addition, we worked with our development team at ways to use the personal information that the user provided on the initial page, to check the customer database and suggest the user might have a profile already if we had a match or partial match to the name, date of birth and address of the customer.
The main achievement of the project was bringing 29 individual 'pages' of the application into three.
1) One page which progressively disclosed questions as they came, tailoring themselves to the users answers to the previous question. Clear sections and headers, inline validation of values, amounts and smart address searching. Want an ISA? We need your NI number. Where can you find that? A payslip might be easiest. Really going through every label, error message, help text with a fine toothcomb to ensure that no one got left behind, baffled by jargon and terminology. Not answering inapplicable questions. Not having to skip between pages and pages of information.
2) One page of confirming this detail back, and allowing edits. Showing the raft of T&Cs, easily explained, Key Information, FSCS warning messages. A clear call to action, the ability to navigate back and start again. Clear signposting of what was going to happen next.
3) Confirmation of opening and vital next steps information. Reminders of login details, new account number confirmation, and clearly displaying how to deposit money (our original plan was to accept Debit Card funding and funding by Direct Debit - however these changes were not able to be included in this iteration). Details of the benefits of becoming a customer - access to Virgin Money lounges and discounts from a range of suppliers across industries.
We transformed what was a longwinded, pain-in-the-ass, 29 page behemoth of a form into three pages. I felt triumphant. However it wasn't over yet!
My next challenge? Rounds of user testing the new form. It was a big job, and I enlisted the help of our specialist Usability lab, who take the seemingly pleasurable job of travelling round our lounges up and down the country and borrowing people's time to take them through the form. Test the limits, break it, and feedback to me to continue iterating the form, rearranging items, sparking ideas for ways to reword things, cut down text, realising we should have smart defaulted another option in a menu, or users in this scenario expect a different question to come next when they are asked the previous one.
One of the more surprising pieces of user feedback was a dropdown product selection. (see below). Users were continually trying to scroll down the page to (we suspect) check the length of it. If they did this whilst the dropdown was highlighted, they ended up applying for an incorrect account, which trust me, if it was an ISA and the user already had an ISA, this created a paper trail so long it could reach the HMRC and back (and often did). So we disabled scrolling on the dropdown, and pulled out the account name and rate on page so it was clearer.
Here is the live form following development.