« View all blog posts

Study: Behavior is more predictive than demographics

We analyzed hundreds of client campaigns to find out what kind of data is most predictive of customer actions. The answer? Behavioral data is 20x more predictive than demographics alone.

If you’re a marketer, chances are you already know that data-driven marketing performs better. 70% of IT executives say data technology contributes to revenue growth. Data-driven companies are six times more likely to be profitable than their competitors. And 75% of companies see an uptick in engagement when they turn to data for their marketing needs.

These stats are pretty compelling. But here’s the question:

What kind of data is driving these results?

Because the truth is that not all data is created equal. And what we marketers need is data that is predictive. Data that tells us who is likely to convert. Data that helps us stop wasting budget on people who aren’t likely to convert and hone in on those who are.

So, what kind of data can do all that?

The answer, according to a study we did behind the scenes across hundreds of client campaigns, is behavioral data.

Past behavior predicts future behavior


To better understand what was working for our clients, here at Lytics we did some digging. We analyzed data models across all Lytics customers, looked at successful campaigns, and asked the question: Which data is most predictive of success?

Was it demographics telling us who was likely to convert? Was it content affinity? Or was it behavior?

The answer? Behavioral scores were the best predictor of people’s next actions.

In fact, 86% of the time past behavior was the leading predictor of future behavior.

That’s right, 86% of the time.

Another 10% of the time, the greatest predictor of future actions was content affinity.

And demographics? They limped in in last place with just 4% of all client campaigns.

But wait, why is behavior so much more predictive?

We think the answer lies with the nature of the data. Demographics are general and static where behavior is specific and real-time.

Demographics can tell you I’m a woman in her mid-30s. Assumptions about that demographic group without any additional data might have you targeting me with ads for things like lipstick (which I don’t wear), engagement rings (which I can’t imagine ever wanting), or weight loss ads (which I find offensive).

Demographics, in other words, are too general to get you to the heart of my passions, my connection with your specific brand, and my buying patterns. They require a lot of guesswork or an additional contextual layer of behavioral data.

My behavior, on the other hand, can tell you that I love hiking and will open any email you send me with a headline about hidden trails or mountaintop views.

Behavior can tell you that I’m a bargain shopper and you can push me over the purchase finish line on a product I’ve been considering by offering me a coupon.

Behavior can tell you that I’ve purchased trail running shoes from the same brand in the same style every 8 – 10 months for the last few years (and they’d be smart to send me a nudge in another 8 months since my new shoes just arrived yesterday and I’ll be hitting the trails hard with them every weekend).

Behavior, in other words, gives you real-time context into what I care about and how I’m interacting with your brand.

So, are demographics worthless?

Now, does this mean demographics don’t matter? We don’t think so. Marketers have used demographics for years because they do have some predictive power. They do tell us something about our audiences.

If you’re marketing Viagra and you have data that shows men are vastly more likely to buy than their lady partners, knowing I’m a woman is probably helpful to you. If you’re targeting people who are about to collect social security, knowing I’m in my mid-30s can help you exclude me.

The point isn’t that demographics are bad or unhelpful. It’s that they aren’t enough to really get you to the heart of a customer. They’re not enough to truly predict future actions.

That’s what our study told us, and it’s what we’ve observed over and over again with our clients’ campaigns.

How do we harness the predictive power of behavior?

So, great. Behavior is 20x more predictive than demographics alone. But how do we track and understand behavioral data? How do we take advantage of the opportunity it provides?

At Lytics, the answer we came up with is called behavioral scoring, which groups users based on their behavior—both in the past and predicted for the future.

Behavioral scoring tells us both how people behave and what the context is around their behavior. For example: Let’s say we have two customers: Jay and Ty. Jay visited our site 10 times and Ty visited five.

If we just look at those two numbers, we’d think Jay was the customer to target. Because the higher the number of visits, the more engaged the customer, right?

But if we find out Ty visited five times in the course of two days and Jay visited 10 times over the course of the year, that changes things. If we find out Jay’s visits were quick and didn’t result in any click-throughs to additional content, but Ty’s visits lasted 10 minutes each, that changes things.

Behavioral scoring is not only a look at people’s behavior on a high level. It’s also the context your data needs to paint the full picture of that customer’s engagement with your brand. It’s built to track behavior across a variety of dimensions to give us a better sense of whether Jay or Ty might be our target customer for a particular campaign, content piece, or ad.

Our built-in, out-of-the-box scores—the ones that were most predictive of future behavior in 86% of all client campaigns—track behavior and context around these six things:

:: Momentum (is a user’s engagement with your brand increasing, decreasing, or leveling out?)

:: Intensity (is their engagement above or below average?)

:: Frequency (how consistent is a user’s behavior over time?)

:: Recency (how recently has a user truly interacted—and not just accidentally clicked through to your site?)

:: Quantity (how often do they engage?)

:: Propensity (based on their lifetime engagement, how likely are they to come back and engage in the next 30 days?)

Behavioral data for your business

There’s lot of good news in the research above.

We know that behavior is highly predictive and drives incredible results for brands. We know that content affinity and demographics both also play a role (and combining the power of all three is your best bet for a great long-term data strategy).

Even better, we know that because behavioral scoring is built into our platform and available to customers right out of the box, most Lytics clients start seeing results—like campaigns that perform 2x, 3x, or even 10x better—within just 60 days.

If that sounds like a big win (it is), we’d love to chat. Download our data science white paper to learn more about behavioral scores and how the data science behind the Lytics scenes works. Or chat with one of our experts.

Fair use statement

Want to use our infographic on your site? Please do! Just make sure you link back to this post with a do-follow link and credit Lytics.