The Power of Real-time User Recommendations

Data Analysts get excited about predictive analytics and other advancements in data mining, statistics, modeling, and machine learning. Marketers are happy when they find tools that help them understand information about the effect of their campaigns or optimizations. Business Execs are giddy when they see that upward slope on their profit dashboards.

Whether you are an Analyst, Marketer, or Executive, one accountability that unites us all is business results. A friend of mine who is a CEO of a media company tells his team: “if it is not your responsibility to sell something, it is your responsibility to support someone who does.” His company does quite well.

Real-time user recommendations based on behavioral data is an advancement we can all stand behind because it unifies us in our quest to improve business results.

Data Driven Next Best Offer

Step for a moment into the proverbial shoes of an “average” internet user, one who does not understand behind-the-scenes online customer behavior tracking or buyer personas. For just a moment, log on to Amazon like your grandmother would. Amazing, isn’t it? Amazon magically knows exactly what you what to buy.

It is not just your grandmother. True story: I spoke with an Ivy League business school graduate who runs a successful online business earlier this week. He told me that he was researching some company that offered a new technology and was then amazed at noticing their advertising everywhere he was. I explained that this was probably due to a re-targeting cookie that had profiled him and the kind of technology he was interested in. He had no idea such things existed.

Stepping back into your own data-driven and KPI-focused shoes, consider the way sites like Amazon know what you want to buy or how Facebook knows which updates matter to you. The illusion of magic is created behind the veil of a form of predictive analytics called “Next Best Offer” (NBO).

NBO is rooted in the historical use of collaborative filtering which enables sites to direct visitors toward an attractive offer based on their purchase history and page visits. Today NBO is the solution to the woes of inbound marketing. It enables businesses to amplify the business results from each digital touchpoint. It empowers companies to understand each customer intrinsically, in order to be able to match them with the offers that are most likely to appeal to them (and improve business results for the company).

According to SAS, the use of predictive NBO leads to a “sustainable competitive advantage” and increases response rates as much as 10X.

A defining element of NBO is that it must happen in real time. For example, at Orange UK, NBO is used by agents to determine, in real-time, the optimal retention strategy for each customer. Agents are automatically presented with all relevant details about the customer, a calculator that shows how much they can spend to retain the customer, and all of the potential products and tariffs that can be bundled into the offer. As a result of implementing this method, the telecom giant increased gross operating margins by £2 million monthly.

Product Affinity Analysis

Built from deep analysis of raw event data of all consumer actions across all channels, product affinity insights give companies something much more important than data. Companies can now get an accurate prediction of whether a user should be shown another recommended product, and if so, which product.
Product Affinity Insights automate answers to complex business questions like:

  • Are users who view a certain product likely to end up buying a different one (perhaps one that is cheaper or has better reviews)?
  • Are users who buy one product more likely to buy another? Which product is more likely to be purchased together with the first?
  • What causes a customer to search and purchase the same product on another site?

This has been mastered effectively by Amazon’s “Customers Who Bought This Item Also Bought” section.

With accurate, data-based answers to these questions, companies can optimize the shopping experience across all channels, quantify demand for complementary items, and provide data-driven information to feed recommendation engines.
For example, in this analysis, that looked at all product pages viewed by customers across all touchpoints (including the retailer’s various websites and apps), we see that viewing the Windows 10 product page is frequently followed by a view of Windows PC product page and vice versa. Clearly these products should be recommended for one another.

product affinity analysis
Product affinity analysis conducted on product pages views across retailers sites.

The Right Suggested Content Promotes Stickiness

A website’s “stickiness,” its ability to keep a visitor’s attention, is measured either in minutes per month or pageview per month. The KPI is an important one for advertisers, who will pay more for sticky sites on which their ad will be displayed more.
For this reason, content publishers often use “recommended content engines” to display related content to encourage users to visit different pages or new categories on the site.
To do this effectively, we collect data on reading habits, track the articles that users read (and their categories), and analyze the information using our behavioral path analysis.

In this example, on a huge publisher site, we performed a behavioral path analysis and found that readers of the business section are the ones that continue to read more content from other sections. They mainly go into the politics section. Some would argue that business and politics are intertwined, the data proves that if not woven together, they are definitely linked. Out of the 70% of visitors who read the business section, 39.6% continue to read the Politics section. 13% actually continue to entertainment, sports, and all the way to tech. This journey shows this most popular path, for people who read more than 2 sections.

Behavioral path analysis
Path analysis of readers behavioral patterns

With these insights we can than promote suggested content from the right categories to promote stickiness. We can see that this is the hardest with sports users (who typically only read the sports section).

The value to the business

A marketer tasked with reducing bounce rate will have an interest in tactics that increase stickiness. But, on a strategic level, these suggestions, based on real-time behavioral profiles, not only reduce bounce rate – they increase advertiser spend, which increases business profits.
As my CEO friend says, those who are not directly selling something need to be directly supporting those who do. Now, with access to this kind of advanced user behavior analytics, across every vertical and in every job position, you can make decisions that impact sales, retention, and stickiness – and create business effects that are everyone’s responsibility and accountability.

Ready to learn how behavioral analysis that feeds your suggestions engine affect your business results? Schedule your demo today

Share this post