Customer Profiles

Customer profiles

Quick Definition: A data file that compiles traits and behaviors of an individual.

Key Takeaways:

The following information was provided during an interview with Nate Smith, group manager of product marketing for Adobe Analytics Cloud.

What are customer profiles?
What are the benefits of customer profiles?
What mistakes do companies make with customer profiles?
How do companies build customer profiles?
What information is captured in a customer profile?

What are customer profiles?

At a basic level, a customer profile is an entry in a database — there's a hashed ID for each person. As that person interacts with a brand or any digital experience, they leave digital footprints all over the place.

Customer profiling takes those footprints and ties them back to that user — whether they are known or anonymous.

Profiles are dynamic. They should be updated in real time. That's the key differentiator from the big concept of what a customer profile is to what it can be now.

As we have more interactions, those footprints are captured, and the profile is dynamically updated so that we can essentially use that profile for marketing, engagement, or personalization purposes.

The idea is to get to a profile that is rich with information so we can engage with people in a personalized way, and even in one-to-one engagements.

The ideal customer profile is unified — the data that is collected about customers from across the enterprise is stitched together.

This means that customers who appear in your CRM, in web analytics, in a customer support database, and on your email list are all known as the same person — their identity is resolved across databases, channels, devices, and internal departments.

The unified profile is a collection of traits and behaviors where we normalize the data that comes in, and it’s then accessed by systems of action, like an email system or an A/B testing solution.

One of the key things about customer profiles is that they’re portable so that we can have consistent cross-channel engagement experiences for consumers.

A profile needs to sit close to systems of action so that whatever it is that's accessing that profile to deliver an experience can do that in real time. It shouldn’t be buried three layers deep in tech — it needs to be a top-level layer that can be accessed.

What are the benefits of customer profiles?

With the nature of how we interact, companies have multiple martech vendors that engage in multiple ways — through email, through display, through social.

With a customer profile, you can have the most up-to-date information on a customer that you can then activate in any channel.

If I have information that comes from a click on an email, and then the customer goes and accesses the website, that information is going to flow to the customer profile.

The next time I advertise that profile, it's going to be more updated and I’m going to be able to offer more personalization.

Let's say that I live in Los Angeles, and I go to I decide that I want to potentially buy a truck. If I start throwing off signals with my behavior on the website — for instance, if I go to the build-a-truck option versus view offers and incentives,

I’m giving off a behavioral signal that I might not be price sensitive, so that could be a good signal for Ford to combine with, say, the data point of “from California” so that the next email that I get shows a truck in the Hollywood Hills. They know that would speak to me.

You can use a profile to catalyze any marketing channel and so instead of wasting time and money with a lot of educated guessing, you can actually use accurate, up-to-date, real-time information in any channel. You can orchestrate complex journeys and you can approach more personalized marketing.

And at a base level, customer profiles help you identify your most valuable customers. When you know that, you can augment or magnify it.

You can do look-alike modeling to acquire more potential customers that look like your high-value customer profiles.

We talk about most profitable customers. A lot of times we equate them to being the most revenue-generating customers. But companies have some revenue-generating customers that cost them a ton of money. That information can also be collected in a customer profile.

What mistakes do companies make with customer profiles?

A lot of companies start off with some kind of a vision of who their target customer should be and then they try and force the data.

It's kind of a long way of going about it. So they start with some kind of description of an ideal customer, their target market, and they’ll include some details, but it’s not a trait database.

They'll include background, hobbies, interests, education, income level, or even if the customers have dogs. It's almost like old-world media buys, old-world database marketing, where everything revolves around a few key demographic things.

And then companies start trying to figure out why their campaigns are or aren’t working on these groups.

And typically it's not working because they’ve over-generalized and made assumptions versus having the data tell them what their customer groups are and how they should be interacting with them.

A lot of organizations think that as long as they can just collect all the data, they'll be good. But there are significant cost and time implications for that.

The way a lot of organizations are trying to tackle customer profiles is that all the data gets dumped into a data lake and then they have data scientists that write SQL queries and try to figure out a model to build customer profiles.

And that usually takes days, if not weeks, if not months — and then you have to update and rationalize the data.

And customer profiles have a shelf life. If someone is walking down the street and walked by a Starbucks, triggering the geofence with their app, Starbucks has a window of maybe two minutes to engage the customer and entice them into that store.

They can't wait for a data science team to write together history profiles.

There's a real-time component that needs to happen. And a lot of organizations think that if they can just centralize the data and have one source of truth, they can successfully build profiles. Technically they can, but these aren’t actionable profiles.

They're not portable profiles when they're needed to actually engage with a customer. And a customer profile layer has to be close to systems of action, not buried three to five layers deep in the data lake with the data frame.

How do companies build customer profiles?

We collect trait and behavioral data from all the different channels that customers interact with — your email system, their demographic information, social interaction, whatever all those interactions are.

There are different categories for those data and they’re typically siloed off, though once you move them into a platform and normalize them, they can be integrated into a single profile.

So that's really key, that you can actually have flexible, holistic data collection from multiple homegrown systems or tech stacks.

You collect all of this data in various systems, and you have it in a data lake or some other central repository.

And then when events happen — whenever someone comes to a website, engages in a mobile app, walks in a store, or hits a geofence — you tie the data to those profiles.

Whenever we recognize a profile on a device or a customer doing something with a digital engagement point, that action is then basically assigned a line item in the data record of their profile.

You could almost think of it as a profile where events are allocated to that profile.

The next piece beyond that is defining what a profile is across channels and devices. For instance, you have to resolve devices to people, because a customer could access their email channel on multiple devices in different ways — that's how we interact as consumers with brands.

Once you connect devices to people, you have to deduplicate and purge certain things, and then you have new data that comes in as well that augments that profile.

And that’s where you're actually going to be able to activate those profiles. At some point, if your system has AI or machine learning built in, it'll automatically group profiles of a similar nature.

You want to create groups — or audience segments — that can potentially come together and scale, because the promise of one-to-one is good, but businesses want to scale as much as they can.

What information is captured in a customer profile?

The most generic level of information is trait-based information — the information that would exist in a CRM platform. For instance, age, demographic information, psychographic information, account status, whether the customer is a gold member or a platinum member.

Profiles get really interesting when you add behavioral information to customer traits.

As you start to understand the interaction of traits and behaviors together, you can make better decisions about where to invest your marketing dollars and how to personalize experiences to different customer segments.

People also view