Use data to predict customer behavior and design better products.
Do you know which customers are most likely to stop using your product in the next month? Or, what actions your best customers take with your product when they start using it?
With the right data, product managers not only know the answers to such questions, but they also know what actions to take to keep customers and a whole lot more.
This is the area of predictive analytics and our guest is Brian Brinkmann, the VP of Products for a company involved in the revolution of business intelligence tools, leading to greater predictive capabilities. That company is Logi.
Brian is the perfect person to learn predictive analytics from because he is also a classic product manager, recognizing the value of customer interactions along with predictive data.
Summary of some concepts discussed for product managers
[1:53] What was your path into product management?
My first job in electrical engineering was in control systems for power plants, which led to a project designing user interfaces for those control systems. I learned about human-computer interaction and how to involve people in the process. From there, I went back to school for a dual degree MBA and Master of Engineering Management. I knew I wanted to go into product management, but needed some experience in the field. I worked as a strategic consultant and then eventually made my way into marketing and product management. My story is proof that you do not need a specific background to get into product management. If you want to do it, you’ll learn the skills you need to be successful.
[8:25] How do analytics figure into your work?
Product managers of applications like CRMs and healthcare management platforms know their business very well but often misunderstand how complicated analytics are. They need to get those analytics into the user experience so that the end users can get the data they need.
[10:22] What kinds of insights are you looking for in analytics?
We are looking to see why things happened and what will happen moving forward. If you can figure out what might happen, you can begin taking actions against it. A financial company wants to flag a fraudulent transaction right away. An iOT company wants to know that a machine failure is coming so they can try to prevent it from happening. It’s also a good way to understand customer acquisition and how to hold on to a customer. It’s much easier to maintain a relationship than it is to start a new one.
[13:38] Can you give an example?
If you are a $50 million per year business and your churn rate is 6 percent, if you can reduce it by half a percent, you’ll save $500,000. Everyone is excited about artificial intelligence and machine analytics, but we advise people to start by determining what their business problems are and what’s the best way to solve them. Otherwise, you are just using technology for technology’s sake. We also work with healthcare organizations to determine how likely someone is to be a no-show for an appointment based on their profile and past behavior. If someone is not likely to show up, they can send a reminder. Businesses can also use predictive analytics to determine if they are overstaffed or understaffed on a given day.
[17:40] How can product managers use predictive analytics to make decisions for their business?
The outcomes are as good as the data use you use to train the models. There might be seasonality involved or other factors. We advise people to monitor their models and track to see how well it did compared to its predictive outcome. You always need to be testing your assumptions and make sure the model is working. You have to be mindful that models will work in certain circumstances but not in others. There are people who will take action based on what those models say, so it’s important to make sure they are as accurate as possible.
[20:43] What role does big data play in predictive analytics?
If you’re looking for predictive indicators, you don’t have all the data you really need to get the answers. For example, you might need weather data or something from the Centers for Disease Control if you’re a healthcare provider. You can combine data from these sources with your own information to drive up the accuracy of your models. The more data you bring in, the more work you need to do to have it make sense. It takes time, but can be worth it in the long run if it makes your model more accurate.
[25:12] How are predictive analytics used in marketing?
Predictive analytics can tell you how likely it is for someone to remain a customer after they enter your system. In the retail banking world, the number of accounts you have defines how long you’ll stick with them. That’s why whenever you open a checking account, they ask you about opening a savings account or a credit account. In any industry, if someone comes in as a lead and you don’t follow up with them, you can lose them.
[27:05] What are some common mistakes people make in this area?
Starting down the analytics path without a clear business objective is one big mistake. If you can’t articulate the problem you’re trying to solve, you are using a tool to find a problem and it’s a waste of resources. We usually have customers come to us with one problem and we find several others along the way. People also underestimate how long it takes to work with data. The data is often not as clean as you thought it would be, which creates problems for algorithms that like clean data. You also can’t assume that a model will live forever. You need to keep an eye on it to make sure that nothing changed along the way. The winning combination comes when you put the power of computers together with the power of humans to analyze what the computer is giving them.
[32:20] What’s the balance between using data and customer experience to make product decisions?
Do we incorporate customer requests or enhance the product based on where the market is going to go? If you only satisfy customer requests, you will probably have a stable business but it won’t grow. If you don’t apply the data in context of customer pain and market trends, you’ll find that the product becomes irrelevant. You can’t just go by the data either, you need the human component to provide context to the data.
“If you think you can or can’t, you’re right.” -Henry Ford, though attribution is debated all the way back to Virgil.
Thank you for being an Everyday Innovator and learning with me from the successes and failures of product innovators, managers, and developers. If you enjoyed the discussion, help out a fellow product manager by sharing it on your favorite social network.