“I’ve invested in analytics and have thousands of data points streaming into our databases, yet life doesn’t seem any easier.” Unfortunately, this is a very common complaint about Big Data investments. Teams get overwhelmed with too much data, and analysis often takes longer if you throw more data at a problem. A newly emerging trend in this space is the use of “prescriptive analytics” — rather than simply collecting data, your team can rely on automated analysis tools to generate recommendations (i.e. prescriptions). This means no more lengthy data reviews, but simply allowing the prescriptive tool to make business decisions on its own.
If this sounds too good to be true (or scary!), you might be surprised to know that many companies already use such approaches. Review the examples below to see how companies are using prescriptive analytics to automate and grow their businesses.
1. Product recommendations on your site and in your email.
When purchasing products online, you’ve likely been recommended related or similar products. This is a useful-yet-simple form of prescriptive analytics: there is no sales rep on the other side wondering who you are and what you’d like. Instead, a set of algorithms determine what to recommend and how.
Indeed, using product recommenders play a crucial role for many companies. Even with a small customer base, you’re likely to increase sales or customer engagement with almost no additional time invested. A challenge could be incorporating product recommenders into more outbound forms of outreach, such as within emails or in your call center.
2. Rule-based discount codes based on future activity.
Suppose you could predict whether a customer is likely to buy within the next 90 days, and that you could estimate how much they’ll likely spend if they do buy again. Giving customized discount codes to individual customers with a high risk of churn (i.e., not likely to buy again) but who have been big spenders could get those individuals to spend quite a bit more, and give them an incentive to return in general.
This process can be automated as well: using a predictive model, your customer insights team can set a business rule that whenever a customer (a) has a bigger expected spend than $100, and (b) is unlikely to buy again, then send them a discount code for $10. Of course, the actual values and amounts depend on your business model. Regardless of values, however, this process can be automated, with models updating and scoring customers, constantly looking for opportunities.
3. Markov chains to predict customer success.
Don’t let the term “Markov chains” scare you — the idea here is simpler than it sounds. Imagine you track which customer activities lead to which other activities — people who read emails have a 20% chance of clicking a link, and people who click links have a 10% change of buying. By “chaining” these probabilities, we can see that someone who gets an email is 2% likely to buy. Taking this a step further, we can then begin to see which activities are likely to convert a visitor, and recommend those activities to them.
Your customers follow specific, often predictable, patterns across your sales channels. It is possible to score each activity to determine how likely a customer is to make a purchase. Did they visit your special “high value customers” page? Did they visit a landing page catering to Cyber Monday deals? If so, you can provide them calls to action that specifically speak to their interests, and oftentimes, this might not mean encouraging a purchase. For example, someone visiting a Cyber Monday page might need to be emailed content to bring them back a week later, rather than focusing on making additional sales.
With this in mind, you can use Markov chains and automated funnel analysis to determine the best steps for each customer, and encourage them with those calls to action.
The tools above all leverage Big Data — they see which customers, web visitors, and leads do what, and how best to encourage them to engage with your company. They’re prescriptive, which means you don’t need to strategize around their findings. A product recommender simply generates recommendations without anyone at your company having to reflect about that specific individual. Not only is this likely to help grow your sales, but might even give your sales reps some breathing room this holiday season!
|Next Post:||Canopy Labs announces analysis of over $1 billion in commerce; releases new dashboard and analytics tools|
|Previous Post:||A Heat Map of Consumer Sentiment For The Past 13 Years|
Click here to subscribe to Canopy Labs Insights!