Data Never Sleeps 7.0 - Image credit: Domo

At a time when customers demand higher quality experiences from brands in exchange for their data and their dollars, this treasure trove of information represents a golden opportunity for marketers to finally achieve that elusive “single view of the customer.” But, given the sheer size, scale, and speed of the data that marketers must handle, traditional and manual data mining and analysis won’t suffice.

Enter artificial intelligence (AI). With its powerful models and algorithms, artificial intelligence can help marketers and organizations of all sizes make sense of the petabytes upon petabytes of data that they are now dealing with. 

In the enterprise space, for instance, entire sectors of business have emerged and evolved to create, capture, store and take action on all of this data. For instance, there are more than 7000 vendors that marketers can consider using in as they build out their marketing strategy and on average in 2019, marketing organizations are using fifteen data sources (up from just ten in 2017). Read on to learn about three ways AI can help you create and act upon a single view of the customer.

1. Resolve identity to reach people, not devices

The foundation for creating a 360-degree customer view is identity. Marketers must first know whom they are speaking to before they can start having a relevant conversation. And that doesn’t just mean customers who have given personally identifiable information (PII). Identity can also include browsing behavior or what devices they are using.

While conceptually determining one’s identity is as easy as matching two data points together (e.g., email address, phone number, loyalty ID), things get tricky as consumers now interact with a brand in more channels than ever before, often not bothering to “raise their digital hand” to become identifiable. When compounded with the fact that nearly two-thirds of customers have used multiple devices to start and complete transactions, you can begin to see the major challenge that marketers face.

The good news is that the rise of AI has brought more powerful algorithms that can recognize patterns even when PII is not available for a direct, deterministic match. This is called probabilistic identity matching, or “fuzzy matching,” and allows marketers to increase the potential reach of their messages, ensuring their content can be more personalized. For example, an individual who has been browsing similar websites and content, from the same IP location, but on different devices, like a phone and a tablet, would normally receive a stock version of a marketing message for each device since their profile is separated. However, by letting AI work it’s magic, within a preferred level of confidence for the match, this individual’s identity can be linked to both devices. That means that we’re now considering the person behind the device and not just the devices themselves, enabling a more personalized offer to the individual wherever they are browsing, using the data collected from all of their devices.

If you are considering how best to resolve identity — deterministically or probabilistically — think about the following:

• Do you want a 100% match? (If so, use deterministic)
• Are you ok with the match being accurate to a high degree of confidence, that you choose? (If so, consider probabilistic or a blend of the two)
• If the match were wrong, would there be any negative consequences to what you are sending? (If so, stick to deterministic)

2. Make progressive profiling, predictive

So, now that you’re confident in the identity of your customers, you’re ready to start sending personalized messages to them. This is where marketers turn to segmentation with many brands and marketing teams executing sophisticated targeting strategies based on customer profile and behavioral attributes.

But how well do you actually know your customer, and are you using all of the data and insights available? The answer for many is no. Not because they don’t want to, but because of constrained resources and the sheer amount of data to scour through at their disposal.

Here, there are two ways that AI can help, the first of which is about using the predictive nature of machine learning models to augment the customer profile. You may already be familiar with the idea of progressive profiling, a technique many marketers use to ask questions and gather additional bits of info over time through explicit customer feedback. But even this approach often leaves gaps as customers simply fail to respond to the email, web, or even in-person surveys that are often used.

Instead, by observing customer interactions and behavior, then applying AI models, marketers can begin to infer and even predict customer behaviors and preferences.

If you feel there are gaps in what you know about your customers or data points you’d like to use to complete the customer profile, consider taking the following steps:

• Start by building a progressive profiling strategy where you periodically send messages not to sell or position anything, but to ask for and gather more data from your customers
• If engagement rates are low with the messages and surveys you send, consider how you could use predicted attributes to fill in the gap — some examples include what type content or products a customer has a strong affinity or their preferred channel of communication

3. Uncover new personas within your audiences

The other way that AI can be used to round out your understanding of your customers leans on even more advanced tools and cluster analysis models, to unlock the rich insights buried in their data. Here, it’s not just that they can add a new attribute to their existing segment to refine it more, marketers are actually discovering entirely new personas within existing audiences — personas that they never thought to look for in the first place. This has major implications for the way a marketer may choose to message to those customers in the context of a campaign. Take the following (fictitious) example:

Northern Trail Outfitters is an outdoor gear and apparel retailer serving the outdoor enthusiast market. They know they have a large segment of avid hikers and target them with relevant hiking content and offers. But when they applied AI clustering analysis, they found distinct personas like Glampers, who appreciate higher-quality gear and creature comforts while on the trail, and Trail Techies, who are interested in the most high-tech, innovative gear. Now they can adjust their outreach strategy to target Glampers and Trail Techies with different campaigns and content as it makes sense, such as highlighting the on-the-go coffee maker for Glampers versus their new GPS watch for the Trail Techies in their next ad campaign.

As you mature your understanding of your customers and segmentation strategies ask yourself the following:

• Are my current audiences and segments broad and large in size? 
• Could there be smaller (but still significant) segments that have related attributes that present a different marketing angle?

If the answer to these questions is yes, you should seek out the help of your data scientists (if you’re fortunate enough to have them), or research purpose-built AI tools that can assist with this analysis and provide you the insights to inform your marketing strategy.

Making sense of big data can be a big mountain to climb, especially without a rockstar data science team, but with the democratization of AI techniques through enterprise-grade tools, you can quickly begin to resolve customer identity, gain a deeper understanding of your audience and take resulting action more effectively than ever. 

To learn more about how big data and AI are transforming the way marketers build relationships and understand their customers through a single view, download the Fifth Edition of the State of Marketing Report and stay up on all of the latest trends.