Building App Personalization That Doesn’t Compromise on User Experience

Patrick Gordon

Principal Product Manager, Jumbo Interactive Limited

people reacted
6 -minute Read,

Posted on May 5, 2021

Learn how Jumbo Interactive uses Amplitude Recommend to create personalized experiences for every user, based on their on-site behavior.

I’m an engineer by trade. Before I accepted the role as Jumbo Interactive’s Principal Product Manager, I was a software architect and front-end developer. For a long time, I didn’t understand how powerful personalization could be. I honestly thought building a great app or website was enough to attract and keep customers. In other words, “build it and they will come.”

But I was wrong in thinking that a good product was the only way to engage customers. They need a reason to keep coming back, and it takes a continuous effort to entice visitors to return after an initial click or sale. Through our experience with a new solution, I would soon come to learn that personalization is the best way to dial up your marketing efforts, increase conversions, and translate to more sales.

Reselling Lottery Products Down Under

Jumbo Interactive runs Oz Lotteries, a platform that resells Australian National Lottery products like Oz Lotto, Powerball, and Lucky Lotteries’ Super and Mega Jackpots. We also offer a syndicate service that allows people to buy shares in lottery pools and resell tickets for popular charity lotteries like Mater Cars for Cancer and Act for Kids. We recently expanded into the international market and now offer a SaaS (software-as-a-service) version of our platform to lottery operators in other countries.

If we can't sustain the interest of people who play our games, we might as well call it a day. Click To Tweet

While our customers are the lotteries themselves, our product development and marketing efforts focus on their end users: the players. No one does lottery management better than us, but if we can’t sustain the interest of people who play the games, we might as well call it a day. So, we optimize our new products and new features to make playing online lotteries more fun and engaging. Our customers’ customers reckon it’s great, and we want them to tell their friends that they’re having a fantastic time with our online lottery offerings.

Personalization Leads to Engagement and Loyalty

Although I was skeptical of personalization, I came to understand it as the next step in building customer engagement. It’s a way of ratcheting up our ability to speak to customers by presenting them with offers that are relevant to their interests and the way they play.

People are very protective of their personal information and greatly concerned about the way companies use their data. At the same time, they want tailor-made experiences and flock to services like Spotify and Netflix that automate the discovery process and recommend content based on their listening and viewing habits. We wanted to offer a similar experience to lottery players, and needed it to feel inviting instead of intrusive.

Before implementing our personalization program, our marketing team cast a wide net. We had a massive cohort of users, and we frequently blasted them with emails. Some people would end up buying more, but our conversion rates could be higher. For our non-personalized on-site placements, our conversion rate was only 1.57%. Although we could see we were failing, we didn’t have the baseline data to understand why.

Building a Personalization Technology Stack

About a year ago, we started to build a personalization technology stack based on customer engagement platform Braze, and digital experience and analytics platform FullStory. The marketing department pushed for Braze as the best way to insert user-specific information into personalized communications. We had already built an in-house marketing automation pipeline, but it wasn’t yielding the conversions we needed.

The team started looking at third-party solutions to make up for our internal lack of expertise. It made more sense to buy existing products than to spend months creating them on our own. After all, our expertise is in selling lottery tickets, not building up a personalization system from scratch.

We began to collect digital experience analytics that helped us understand how users navigate our products, and that was an eye-opener. We thought we knew what was drawing eyeballs, but when we looked at the actual numbers and how our users navigated the digital experience, we saw that our hunches were wrong.

With Braze, we emailed automated lottery recommendations after somebody made a purchase. But it wasn’t enough. Our conversions were up, but we weren’t hitting the numbers we wanted.

That’s when we learned about Amplitude Recommend. We had started to use Amplitude Analytics as a digital product analytics and growth tool, but this new product was just what we needed to push our personalization efforts to the next level.

I staked out the competition for personalization solutions, and found that Amplitude was the best game in town. The only other company using an equally complex AI-driven recommendation engine was Amazon, but their software is clunky and confusing.

Amplitude Recommend allows us to create cohorts based on users’ on-site behaviors. It then generates personalized recommendations and messages encouraging them to take further actions. Its machine learning algorithms can predict what Oz Lotteries customers will buy next and steer them in the right direction to purchase. We could never have developed something like this in-house, nor could we have implemented it so easily on our own. Amplitude’s concierge team took care of the setup in a matter of weeks.

Experimenting with Personalization

When we rolled out Amplitude Recommend, we started experimenting with our personalization efforts. One of our earliest initiatives was an email campaign that directed charity lottery purchasers to our online shopping cart instead of sending them to the play page. We guessed that presenting them with the option to buy a pre-selected ticket would be interesting to them, but we were wrong. It turns out that charity lottery buyers spend a lot of time on the play pages, looking at the homes and vehicles being raffled. Based on this information, we had to rethink that campaign. That’s the beauty of having insights—you can easily test out ideas with Amplitude and see what works.

Next, we sent out an Amazon-style charity lottery recommendation to users who had made an online purchase. It comprised a carousel-style listing of four similar lotteries a customer might enjoy. We also sent out a push notification to players who had placed an order in the online app with just one recommendation. We then A/B tested the conversion rates in Braze to refine these offers. Finally, we created a content card for people who clicked through an email or push notification to access the website. It’s a pop-up inbox-type thing that lives on our site and creates another touchpoint if the customer hasn’t already converted.

The results have been encouraging: We’ve seen that the email has a 20% improvement over control; the app push has a 32% improvement over control; and the content card has an 8.5% improvement over control.

A 158% Improvement In Conversions

Amplitude Recommend helped us see that every step along the customer journey is an opportunity to serve personalized content to increase sales and make playing the lottery more fun. Applying these lessons was a team effort: the developers worked on the application stack, I handled the product end, and our Marketing Automation Specialist, Melissa Cowan, dealt with the marketing side.

Once we started offering AI-driven recommendations, we experienced a 158% lift in conversions on one checkout page in two months. That has the potential to translate to an extra $500,000 in sales year over year, and it’s only the beginning. I’ve seen the difference that personalization can make, and now I’m a believer.

Armed with Amplitude Recommend, we can start to improve the core experience for different customer cohorts. I’m especially interested in targeting the 20% of customers who spend more on an advanced playing experience. These are people who play the same numbers every week or analyze previous draws to create combinations that can beat the odds. I’d like to find ways to enhance their digital experience, and I can do that by understanding what draws them to the site and keeps them coming back.

Transforming a Black Box into a Tool

Amplitude Recommend transforms the black box that is machine learning into an intuitive tool that provides real-time, actionable analytics that power product and marketing decisions. It integrates seamlessly with Jumbo Interactive’s technology stack and amplifies our ability to serve relevant and engaging content to Oz Lotteries customers.

With app personalization, aim to balance user experience with your optimization goals.

Some companies choose optimization at the expense of user experience. They over-optimize to drive revenue, but it scares people away. With Recommend, we’ve found the perfect balance between optimizing revenue and creating fun for our customers. You know when you’re listening to a Spotify playlist and think, “Wow, Spotify just gets me.”? That’s the feeling we hope to create with Recommend, using a fraction of the effort required by other tools.

We’ve moved past the growing pains and have gotten the hang of tweaking and optimizing our offers. It’s time to ramp things up. I want to offer better products, increase customer engagement, and boost the social and interactive component of our online offering. I also want to share everything we’ve learned at Oz Lotteries with lottery operators around the world. But mostly, I want people to have more fun. After all, that’s what playing the lottery is all about.


Learn more by exploring Amplitude Recommend with a hands-on demo.

Patrick Gordon

Patrick Gordon is the principal product manager at Jumbo Interactive Limited, the leading digital retailer of both national jackpot lotteries and charity lotteries in Australia.

More from Patrick

Inside Amplitude