No one would disagree that once a reader is interested in buying a subscription, it is quite a pity to lose him just because it is too complicated to buy. This case study is about how the team at Aftenposten worked on such an issue and what we did to drastically improve the conversion rate in our purchase flow.

Challenge

Our starting point was that approximately 17% of the users who entered the purchase flow managed to buy a digital subscription. But what happened to the rest of them? Let’s have a closer look.

To buy an Aftenposten subscription, a user needs to be logged in to the Schibsted Payment Identification system (SPiD) — the common log-in solution for companies owned by Schibsted. Our main hypothesis was we lost sales because we force users to log in before paying for a new subscription.

Users were grouped into two categories: those who were logged in with a SPiD account and those who were not logged in with a SPiD account. The important thing to notice here is 85% of users are not logged in when trying to make a purchase. We had a theory upfront that users not logged in would experience greater difficulties carrying out a purchase compared to users who are logged in.

While 48% of logged in users ended up completing the purchase of a digital subscription, only 1.6% of users who were not logged in could complete the purchase. Analysing all steps of the purchase funnel further, we saw 98% of users in the funnel dropped out before they were able to log in.

The number of users not logged in who were able to complete a purchase was drastically lower than what we initially expected.

For logged in users, completing a purchase was not a problem, as they were already logged in. However, for users not logged in, the log-in process before purchasing caused a considerable hurdle.

Analysing the numbers further and running user tests showed most users simply do not want to spend time figuring out their username and password. Most of them experienced one of these three scenarios:

1. Do not remember their username or password in SPiD.
2. Do not know they already have a SPiD account and try to create a new one.
3. Do not have a SPiD account.

To improve this situation, a new project was set up with the main goal of improving conversion rates for users not logged in.

Guiding users

After identifying the main reasons for users not completing a purchase, we took a closer look at the main components in our subscription system, Siebel, and data we have access to stored in SPiD. Combining these two things gave us an opportunity we had not yet capitalised on: they consist of a lot of user data.

SPiD handles user accounts and payment information, while Siebel stores contact and subscription information on nearly every Norwegian resident. These data are bought from our external partner Bisnode and are regularly synced so they are up-to-date and relevant.

We asked ourselves: Could we build a new solution relying on the data stored to better guide and make decisions for the user through the purchase flow?

Based on this question, we designed a completely new flow. Instead of asking the user to figure out if he has an account and what his username and password is, we designed the new solution to answer the question for the user, based on the information the user already knows (such as his phone number or e-mail address).

When not logged in, users enter the purchase flow and we only ask them to type in their e-mail address.

Based on the e-mail address, we do an automated server side check to SPiD and Siebel, verifying if an account already exists and what kind of information is stored on this account. By utilising the data stored in the systems, we can find users without having the users themselves logging in to SPiD.

Moving payment before log-in

Based on the result from SPiD, the system performs several automated tasks to guide the user through the purchase flow. Depending on the amount of information stored on a given user’s profile, he is sent through different flow processes. The more information stored on the profile, the easier and shorter the flow is that we send the user through.

If the user is found and known in SPiD and Siebel, a new check is performed to verify if a mobile phone number and a valid payment option are stored on the user record. If both a phone number and payment option are available, we send a verification code to the user by SMS. Those who do not have a cellphone number stored receive this code by e-mail.

Upon entering the code, the user is presented with a list of stored payment methods. Selecting the desired payment method and hitting the “pay” button completes the purchase flow. Everything else is handled on the server side, and the user is greeted with a confirmation.

The whole process removes the barriers for the user, allowing her to pay in a secure manner and with no need to log in.

The users are asked to log in on the receipt page after buying a subscription. Naturally, the motivation to log in is much higher to access the content the user has just paid for than to log in to pay. For those who do not remember their password, we offer an option to receive a password-less log-in link by email. We see that more than 60% log in after buying a subscription, instead of 98% dropping off when we force a login before paying.

Many users have a SPiD account, but with no digital payment method stored. For this group, our purchase flow utilises the password-less logging function, where a unique URL is sent to the user’s e-mail address. Clicking the URL automatically logs the user in to SPiD, where he can add then add the preferred payment option and complete the purchase flow.

Compared to a forced log-in, this flow performs much better, but it is still not optimal since it asks the user to leave the checkout flow and check his e-mail to continue.

In cases where users are not found by e-mail and a mobile phone number, we ask them to provide their first and last name and automatically create a new SPiD account for them.

The numbers

Since the launch of the new flow, we have observed a significant increase in conversion rate, first as an A/B-test against the old solution and then rolling it out for 100% of the traffic. Based on data collected over several months, we can see for users not logged in, we tracked an improvement from 1.6% to 8.11% completing a purchase. This equals a conversion rate increase by 6.5 percentage points, or more than 400%. We have also seen an increase in sales from mobile devices, which have increased from 0.82% to 7.26%.

The main purpose of the project was to improve the conversion flow for users who were not logged in, but we also worked on the total user experience. As a result, we saw an improvement in conversions for logged in users increase from 48% to 65%. For all user cases, the average conversion rate went from 17% with the old solution to almost 30%.

As a note, it is important to realise that, when comparing different time periods, conversion can be affected by other factors than only the move to a new flow. However, the increase in sales is too significant to only be attributed to articles, campaigns, and seasonal changes.

With the launch of the new solution and looking at the sales results, we can confirm the original hypothesis stands ground. The results indicate that letting users pay before logging in in combination with guiding users instead of asking them things they don’t know will increase the number of sales.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .