How would you market to your customers if you could predict the future?
We’ve asked this question to many of our customers and we’ve received different answers from almost every one of them. Some would only market to people most likely to buy. Others would spend their efforts engaging those that had the potential to buy, but needed that extra push. Still others would focus their efforts on retaining the customers they already have.
The truth is, there is no right answer. Every business is different and they all have unique goals, but the one thing they have in common is the need to create opportunities to engage and delight their users so that they become new customers or stay happy loyal ones.
Goedle helps companies do this by letting them set their own unique goals and then accurately predicting the likelihood of every individual user to achieve those goals.
While almost every company has a unique set of goals and strategies, they are all very similar in one way; they all want to understand how it’s possible for us to so accurately predict these future outcomes.
User Events and Higher Order Features
We begin with the events that every company tracks. If you have a website, or an app, you’re likely already tracking some basic usage analytics. With this data, we examine how every user of your application behaves; how often they use certain features, how often they visit, and so on.
From this base we begin to layer in higher order features — these features can be as simple as observing the ratio of one event occurrence to another, the city a user is in, or the time of day that a user is active. In other cases, these features can be as complex as the weather forecast in a user’s city for a given day.
Patterns In the Data
Once we’ve begun observing these patterns, we start looking at the users who you’ve told us are your most successful customers, (the ones who’ve recently bought the fuzzy bunny slippers, the ones who haven’t canceled their subscription, the ones who play your game every day), and we begin to learn. Over time, generally about 2–3 weeks, we learn what distinguishes these successful users from others and can predict whether a new user will be successful with a high degree of certainty.
One thing we can’t predict is what’s important to your business, so we make it easy for you to tell us what outcomes are most important to you.
It’s important to clearly specify your goals as this helps us to target the precise users who matter to you.
The Prediction Flow
The prediction flow interface helps marketing teams to understand at a glance how likely users are to achieve a goal. Goedle automatically segments users based on their likelihood to achieve a specific outcome and allows marketers to engage those specific segments with recurring messaging campaigns that increases the likelihood of achieving the desired results.
Importantly, this interface lets marketers understand how their marketing efforts impact user behavior, leading to more effective marketing campaigns over time.
Many customers ask us why we default to showing 7 days of history in the predictions interface. In some cases, like when we’re observing repeat purchases that have a tendency to re-occur every 4–6 weeks, the observation and engagement window is much larger, sometimes spanning months. In most cases however, the actions a user does or does not take in their first week are highly illustrative of their future behavior. This is especially true in the mobile world where apps can quickly fall into irrelevance as users stop engaging with them.
One of our customers operates a global ride sharing app. To encourage users to try the product, the first ride a customer takes is heavily subsidized. Users who take a second, generally unsubsidized, ride have a much higher likelihood of becoming repeat, high LTV customers. Goedle predicts which customers who have successfully completed a first ride are likely to take a second one.
By mapping our predictions to a timeline, we give the marketer an opportunity to engage with their customers and observe how those engagements impact their users over time. We enable marketers to set up recurring engagement campaigns to target individual user segments with emails, push notifications, or other engagement tools. If you have an existing email or push provider, goedle supports many integrations such as SendGrid & Intercom, and we’re constantly adding more to the platform.
Some Examples of What Our Customers are Predicting
- The likelihood of a user to take a second, unsubsidized, taxi ride. This helps drive down marketing spend and increase the targeting of messaging meant to nudge the user into taking another ride and becoming a profitable customer
- The likelihood of a user canceling their recurring subscription. This helps their customer support team focus on users who are at risk of canceling, increasing customer retention and decreasing spend
- Determining whether a user is getting bored with a game, which allows our customer to increase the difficulty of game levels. This increases user retention and engagement, two important metrics for business with ad-based revenue streams
If you’d like to learn more about how goedle can help your business, sign up now.