Back to Blog

Step-by-Step ML: Powering Your App Growth

Jack Riewe

First of all, we’d like to say thank you for visiting us in the new year. This is our first post of 2020 and as you can guess, we plan on keeping our promise to deliver machine learning content focused on the future of work, data literacy, and creativity. This post in particular will focus on using predictive analytics creatively for your SaaS product. The idea for this post didn’t come from thin air.

If you’re a newsletter subscriber, you received an email from us listing out our top posts of 2019. We conducted some data analytics on our web pages and found out our popular blog posts for no-coders as well as our popular case studies. The case study that came in number 1 was about Predictive Analytics for SaaS.

There are countless ways to add machine learning into your app development. As listed in the case study, machine learning can predict conversion KPIs, churn, in-app purchases, build personas, and prevent affiliate fraud. That’s only the tip of the ML iceberg (I’ve been wanting to use that saying). This post will be building off that case study and show you where and how you can insert ML into your app growth.

Let’s get into it.

Step 1: Machine Learning in App Ideation

Besides branding, ideation is your the most creative step in developing your app. Before launching, you want to get a lay of the competitive landscape and identify user needs and pain points. As usual, the ML process starts with data.

The data you want to analyze or make predictions from may include app store data to identify genre gaps and how competitive your space will be. Other valuable data points may include CrunchBase data, review data, language availability data, version history data. You can even go as far as app icon data to begin branding. The purpose of collecting this data is to discover what mistakes your competitors are making, what your potential users are looking for in an app, what will be the complexity of your potential app?

Say you’re looking to build a music streaming app to compete with Apple Music and Spotify.

You would take a dataset kind of like this:


Here you analyze how many music apps are in the app store, see what factors are related to high user ratings, and identify the average price of music apps. You can even predict how an app with similar characteristics will perform.

You can even merge multiple datasets to get more accurate analytics. Just remember, with (accurate) data, anything is possible to predict.

Since we’re a no-code platform, we won’t give you coding advice on how to build predictive algorithms. It’s much simpler than that. You can just upload data, ask questions, and get outputs without the traditional ML process.

Types of Questions You Can Ask:

  • Average price of music apps?
  • Min/Max of size bytes?
  • What is the probability of rating counts for music apps?
  • What music apps are greater than, equal to 5 language numbers?

Step 2: Machine Learning in App Designing/Development

User testing is a vital, vital, vital, part of building your app. We suggest the best way to gather data is user surveys in numerical values. If you can get your app page features, button positions, or copy in numerical value, you can see what factors are related with customer behavior.

For Example:

  • Friendly CTA copy = 1, Direct CTA Copy = 2
  • Button positioned at the bottom of the screen = 1, Button positioned in the top right = 2
  • Number the pages to see what number a user exits the app.

These are just some of the ways you can get insight quickly from machine learning and predict customer behavior before you launch. If you have this data, some of the questions you can ask your dataset include:

  • What CTA copy converted the best?
  • What button position is best for in-app purchases?
  • What page number did the user spend the most time on?

The type of data you collect isn’t just limited to UX. You can also rate overall satisfaction in numerical value or usefulness to get an idea of the overall effectiveness of your app.

Step 3:  Launching Your App With Machine Learning

Marketing your app is the final step in making a successful app. With user data, you can begin to build personas and identify possibly high value customers and build your marketing resources around them.

For example: Once you gather user data, you can begin to segment customers and personalize their messaging differently for increased engagement and brand loyalty. Additionally, you can also predict what types of customers will spend the most time on your app and optimize the UX for them.

With Obviously AI, you can even automate updating the dataset to stay up to date with customer behavior.


Strategically using your data to predict customer behavior will cause your app to grow exponentially as your app and product marketing will be optimized for your users.

Now You Don’t Have an Excuse Not To Use ML When Creating an App

You have the tools to insert machine learning into your app development and a step-by-step guide to creatively predict user behavior to optimize.

If you want to put this guide to the test, request a demo to talk to our CEO directly about using ML for your SaaS product.

Share on social media 

More from the Blog

Predicting Brazil’s Wildfire Patterns in 2020

Can machine learning predict wildfires? We use our platform to ask historical data questions on wildfire patterns in Brazil.

Read Story

5 No-Code Data-Science Predictions for the 2020s

As we enter the next decade, the way we approach data science will change. Here our 5 predictions on how that will happen.

Read Story

Workforce Planning in 2020 With Predictive Analytics

For this use case we will divide how to use machine learning for labor planning into 2 sections: Improving Employee Experience and Optimizing the Hiring Process

Read Story

Never miss a minute.

We will never share your email address with third parties.