Once in a while Google Cloud comes across products which are using their services in the best possible ways. And Google Cloud decides to write a case study about how the company is doing this.

This is how the below case study created by Google Cloud's team surfaced.

Apxor: Helping apps improve their user experiences by using data to understand user’s intent

Have you ever deleted an app you've just downloaded because it was too complicated to navigate? You're not alone. Prabhu Konchada, Senior Product Manager at Apxor, says that users typically give apps between 30 to 60 seconds before they decide if they want to keep it. This is why it's important that app developers understand exactly what users want so they can show them how to get it quickly.

From new feature adoption to creating personalized app experiences, the team at Apxor handles 1.5 billion data points daily. It helps mobile app companies develop nudges that improve app retention, such as tooltips, coach marks, and walkthroughs that teach new users how to navigate an app, introduce new features, or to encourage a new habit on an app. Apxor also takes pride in its out-of-box solution, which allows customers to implement nudges within minutes, without requiring any coding. All customers have to do is integrate its software development kit (SDK) onto their app, and Apxor does the rest.

Today, Apxor's database is fully hosted on Google Cloud with its database on Compute Engine. It uses Dataflow to process user data and TensorFlow for machine learning (ML) intelligence that helps give customers insight about their users.

Apxor also works with Google Premier Partner Searce to continually learn about the latest insights and product developments and ensure that it is always up to date.

"During the [Google Cloud for Startups] program, we were given the opportunity to experiment with different Google Cloud products and had a lot of support from the team. We were truly impressed with the amount of scale that Google Cloud products enabled, which gave us tremendous confidence."

—Prabhu Konchada, Senior Product Manager, Apxor

Discovering the potential of Google Cloud with Google Cloud for Startups

Previously on a different cloud provider, the team at Apxor decided to move to Google Cloud for better bandwidth after exploring a range of products at the Google Cloud for Startups program.

"During the program, we were given the opportunity to experiment with different Google Cloud products and had a lot of support from the team. We were truly impressed with the amount of scale that Google Cloud products enabled, which gave us tremendous confidence," says Prabhu.

Apxor ran a few proof of concepts to determine if Google Cloud was the best solution before deciding to migrate. "We were struggling with a few deployments due to infrastructure before, but with Google Cloud and Compute Engine, we no longer face any infrastructure problems."

Since the switch, Apxor also saw 30% cost savings in infrastructure, with a significant amount of DevOps time freed up due to the managed services that Google Cloud offers. This allows the DevOps team to focus on new product road maps instead of worrying about infrastructure management. As a startup, having this time freed up is invaluable because it allows a very lean team to experiment with new ideas and expand its service offerings.

"We process tons of data every day, and Dataflow and Memorystore are very important ingredients in our entire process because they make sure that these large amounts of data are transferred seamlessly, so that we receive accurate information in real time."

—Kiran Sunkari, Senior Engineer, Apxor

Improving app discoverability with a seamless data pipeline

User experience is crucial in ensuring that users remain loyal to an app. With that in mind, Apxor personalizes 50 million app experiences daily. From the moment a user opens the app, Apxor uses Dataflow as its data pipeline and Memorystore to capture each step of the user journey, including the way a user interacts with an app. It studies each navigation, such as whether a specific user prefers to tap or swipe, and combines that information with survey responses where available.

"We process tons of data every day, and Dataflow and Memorystore are very important ingredients in our entire process because they make sure that these large amounts of data are transferred seamlessly, so that we receive accurate information in real time," says Kiran Sunkari, Senior Engineer at Apxor.

With Memorystore, Apxor can seamlessly serve more than two million requests per second. This multidimensional data is then analyzed on TensorFlow to generate insights, so the next time the same user opens the app, they would be given nudges to enhance the way they use the app. Of course, customers can choose if they want to personalize an app experience on a group level, instead of an individual user level.

Referencing a music app that Apxor had worked with to illustrate, Prabhu says that 50% of users uninstall the app on the same day they install it. To decrease this number, Apxor needed to find the cause of it by identifying user pet peeves.

Based on the data it gathered, Apxor found that users were frustrated when they could not find the tunes they wanted quickly. The solution? To create pop up play and pause buttons to nudge users towards the search bar for them to type the songs they want.

Enabling a positive experience for fantasy gamers during India's biggest sporting event of the year

Apxor was also given the opportunity to work with India's biggest fantasy gaming platforms during the country's largest event of the year, the Indian Premier League (IPL). "During the IPL, we had to increase our compute capacity by three times," says Kiran. "It was the first time we were handling such a huge load, but it was a seamless experience and we didn't experience any downtime."

With Google Cloud, the team could easily replicate its entire setup across different regions and was able to scale up quickly to ensure there was no disruption to the app experience throughout the season.

Support during COVID-19

When the pandemic hit India's shores, many of Apxor's customers delayed their payments, which put a strain on the startup's cash flow. Understanding the situation, the Google Cloud team extended help by providing Apxor with the support they needed, while ensuring that the platform remained optimized.

"It was a very difficult time for us, and the help we received from Google Cloud extends beyond business. It was really emotional and is something we are always grateful for," says Apxor CEO, Krishna Kiran Kasturi. Likewise, the Searce team also relaxed payments during this period, easing financial strain during the pandemic.

"Having seen the successes of Google Cloud products, we are now experimenting with BigQuery to manage really complex queries, and building the ML side of the business with TensorFlow. With the integration capabilities of Google Cloud products, I believe that we will be able to go a long way."

—Prabhu Konchada, Senior Product Manager, Apxor

Exploring new possibilities with Google Cloud

Moving forward, the team at Apxor is busy working on improving its product intelligence. "We are actively working on client side app intelligence, as well as the personalization and recommendation capabilities that we offer," explains Prabhu. An example would be for apps to automatically deploy the right kind of nudge based on a user's journey.

Apxor is also looking into more complex queries to provide more powerful solutions for customers. "Having seen the successes of Google Cloud products, we are now experimenting with BigQuery to manage really complex queries and building the ML side of the business with TensorFlow. With the integration capabilities of Google Cloud products, I believe that we will be able to go a long way."

Here's the original article published by Google Cloud.