Could adding predictive analytics be the silver bullet that launches your application to sales success? Perhaps, if it can be added at the right price point, and if it will enhance existing functionality. Of course you’ll need to consider the returns offered by predictive analytics in terms of cost, increased scope, and impact on your application’s UX.
Predictive analytics has the potential to damage the reputation of your product, if the information generated is inaccurate or difficult to interpret. To help you make the right decision when it comes to adding predictive analytics to your product, consider these use cases.
Scoring-Based Analytics for People-Focused Business Outcomes
Some of the most simple and direct forms of predictive analytics use a scoring system to determine the likelihood of a certain event happening. For example, Salesforce and other popular CRMs use a lead scoring engine that helps sales staff identify the best opportunities for prospecting. This allows reps to focus their time on the most likely chances for conversion, making them more efficient closers.
Similarly, a customer service app could identify accounts that have incidents correlated to lower satisfaction rates. With this data, the customer reps could preemptively contact the accounts to see if everything is going satisfactorily, rather than waiting for a likely complaint.
Score-based analytics that focus on people’s likely future behavior tend to work well within apps for several reasons. First, they report across a spectrum, allowing users to understand that something may be likely to happen but it is not guaranteed. Second, app creators have a choice for where in the UI they want to expose this functionality. The hypothetical customer service app, for instance, could include predictive scoring as an additional feature, rather than within the main app dashboard.
These factors lend clarity and a high degree of flexibility to this form of predictive analytics and make it a good fit for many applications.
Predictive Algorithms for Optimal Content or Audiences
We are all familiar with the type of predictive analytics that is responsible for displaying the content we see throughout the internet. Social media, for example, auto-populates newsfeed dashboards based on the user’s likelihood of finding certain content interesting or relevant. Suggestion tools on streaming music and video services similarly aim to curate content choices for users.
In the business world, such a feature may automatically assign ad content to an app end user or automate sending of marketing materials to a sales lead. Or, it can apply lookalike modeling to identify new lead opportunities with traits similar to existing customers.
But this type of predictive analytics can hurt the end user experience when it fails to function as planned. Facebook recently overhauled their newsfeed algorithm, for instance, in response to people lessening their engagement with the platform after seeing too much content unrelated to their friend groups. A newly launched Netflix rating system also met with complaints because users said it offered too little granular control over the type of content they got to see.
The easiest way to avoid these problems is to downplay the role the selection algorithm has in defining the app’s overall experience. For instance, if you have a content-focused app, offering a small panel for “suggested topics” alongside a typical main dashboard can improve user experience.
Users should also be able to understand how their actions may lead to different suggestions. Recently, financial institutions have started to avoid lookalike modeling for social media ad platforms because the black box nature of the algorithm has them worried about compliance.
Other Suggestion-Based Predictive Algorithms
There are many other use cases for predictive analytics in apps. Many are used to suggest optimizations to end users, such as a social media management app that suggests optimal times to post for maximum engagement.
These features work best when offered on an opt-in basis. Users can request the app to crunch the numbers for them, rather than being bombarded with algorithm controlled suggestions. They can also opt to see opportunities displayed inside of reporting or data visualizations rather than suggested out of context. Most important, you’ll want to avoid implementing a predictive engine that takes control of the app experience away from the user.
Ultimately, the goal of any application designer is to ensure their end users have a good experience, can accomplish their needed tasks, and don’t feel out of control when interacting with the application. If you can meet these criteria while implementing advanced predictive features, then your app may have a winning formula for success.