How data fuels the evolving customer journey

Join online with today’s leading executives at the Data Summit on March 9th. Register here.

This article was contributed by David Majorman, Managing Partner and co-founder of Differential Ventures.

Have you ever had one of those weird experiences where you’re talking about something – a product or service like skinny jeans or a car rental – and then ads start appearing on Amazon, Google and Facebook related to your conversation? No, the big tech doesn’t listen to you, or at least doesn’t listen all the time. You may not realize it, but your communications often lag behind breadcrumbs – in your email, search history, credit card purchases, and other places within the digital universe. Whether it’s a third-party cookie or a data sharing system between technology companies, it’s easy for advertisers and ecommerce sites to design targeted ads that are relevant to your daily life.

Until recently, consumer travel has been heavily influenced by this type of data – breadcrumbs – usually without explicit consent (we don’t count 100-page legal documents pretending to read when we start using the website). Now, stricter rules, such as Europe’s GDPR and California’s CCPA, are changing how companies obtain, use and share customer information. Consumer attitudes are also changing. More and more consumers are closing in on the feeling of how their personal information is reaching them.

To understand the problem, as well as possible solutions, it helps to understand how Artificial Intelligence (AI), powered by machine learning (ML) models fueled by customer data, is deployed to influence advertising decisions throughout the customer’s journey. Modeling behavior consists of two main components: classifying customers into categories and modeling the collection behavior of people in those categories.

How we classify

Humans are often categorized by demographics, including interests based on age, gender, religion, income and marital status, as well as a person’s education, job title, known hobbies, etc. Many of these factors can be determined or inferred from online data, including LinkedIn profiles, public social media posts, and public (and not-public) databases. For AI / ML-powered advertising systems, the universe of potential customers is a dot in a multi-dimensional space, where each dimension has its own weight. Using all the data they can collect about an individual consumer, digital advertising companies categorize each customer up to a point in that space.

Advertisers cannot model everyone individually. Instead, they model behavior based on different factor groups. Each factor can lead to predictions on its own. Suppose you have a sweater that you want to advertise to customers. The data-driven model can determine that a person between the ages of 40 and 45 has a 25% chance of requesting a sweater, which is reduced to 15% if the person is a woman. In fact, Jane Smith, a 42-year-old woman living in Calais, Maine, may have a 45% chance of wanting a sweater, while her 43-year-old husband, John, may have no interest.

Advertisers describe and model people according to hundreds of factors that try to predict buying behaviors and outcomes. Their goal is to spend their advertising dollars to place ads in front of people who are likely to buy products or services. These data-driven factor models, trained using ML and AI, help them do it with considerable accuracy. But times are changing.

Consumer travel is undergoing major changes

One big change that is happening across the Internet, and especially for advertisers, is that regulators are becoming more restrictive about how consumer data can be shared without explicit permission. Therefore, when you email your friends about your upcoming ski trip and start making reservations online, you are less likely to be shown ads for ski equipment. This is because your email providers, credit card companies and other online vendors no longer freely share the information they collect about you with third parties, which these third parties often sell to advertisers.

This change affects online targeted advertising in two ways. First, the change makes it difficult for advertisers to model customer behavior. They already have the model and that model will continue to work for a while. Over time, they will get older and without new data to train them, they will become less effective. Second, it is now difficult for advertisers to describe their customers in terms of their multi-factor model. Many websites have previously used third-party cookies and other metadata from your browser to determine your identity or at least link you to some of your previous online behavior. Under the new rules and changes to browsers, it is now difficult for sites you have not logged in to identify you and model your behavior based on that information.

Today, more than ever, consumers have control over their online customer travel. More sites will encourage them to opt for data sharing, either by paying them for their data or by giving them access to better deals or more online functionality in exchange for their information. However, if the majority of customers dislike it, the ability to model customer behavior and target marketing will diminish, even for customers who choose.

If you like things the way they are, don’t worry. A variety of startups are developing and releasing new products to help ecommerce sites and advertisers navigate this changing world. Some are providing tools to help encourage more customers to opt for data sharing. Introducing new technologies for removing other third-party cookies. In some ways, the problem of protecting the privacy of online consumer data is a cat-and-mouse game. The new rules will temporarily protect rats, but cats will try to find ways to catch their prey. Ultimately, the regulatory environment for the protection of consumer data generally leads to a better environment for consumers.

Companies that use customer travel data should adopt these new solutions for targeted advertising that comply with new rules for data privacy and customer preferences. Companies like Konnecto encourage customers to share data with an opt-in strategy and then combine brands with customer information based on the models trained on that data. Startups like Kahuna analyze a customer’s online behavior to help target marketing without the need for third-party cookies or other controlled private data. Companies should look for ways to encourage customers to agree to share their data for targeted ads using direct returns to opt for reward systems, discounts or data sharing. Most importantly, companies should start implementing solutions now, even though third-party cookies are available in most browsers and the rules are still in the early stages. Any data-driven solution will require a complex set of data to train ML-powered systems, and if competitors collect enough data first, they will have a competitive advantage.

The message from consumers and regulators is clear. People should have the ability to dial up and down the availability of their data. For those who share their information, advertisers and ecommerce sites should do a better job of presenting these customers with the products they want to advertise, hopefully with discounts and other benefits. People who value their privacy over a more targeted customer experience should see more general ads, or not at all if they choose to dislike and block ads. This evolution may take time, but as many governments transfer power to those who generate data online, progress in a better, more private, and breadcrumb-free digital universe seems inevitable.

David Majorman is the co-founder and managing partner of Differential Ventures.


Welcome to the VentureBeat community!

DataDecisionMakers is a place where experts, including tech people working on data, can share data-related insights and innovations.

If you would like to read about the latest ideas and latest information, best practices and the future of data and data tech, join us at DataDecisionMakers.

You might even consider contributing to your own article!

Read more from DataDecisionMakers

Similar Posts

Leave a Reply

Your email address will not be published.