Photo illustration showing the power of predictive analytics

How to Use Predictive Analytics For Better Marketing

Stop drowning in data. Predictive analytics is a power tool that can turn your data overwhelm into data mastery - and get you up to 3000% better results.
Article Outline

Predictive analytics isn’t scary, isn’t confusing, and isn’t here to take your job. It’s a power tool to help you deliver dramatically better results.

Uncertainty crimps business. It makes it harder to invest, harder to prepare, harder to know what to focus on.

This, of course, is part of the human condition. If we knew the future, things would be easier, right? At least for business.

While AI cannot tell us the future (yet), the algorithms are smart enough to make predictions. These predictions may not be perfect, but they remove at least some of the uncertainty from business. Just being 10% or 20% more accurate with our predictions can result in millions to tens of millions of extra revenue every year.

For marketers, predictive analytics can be a game changer. It can give us clues about which customers and prospects to invest in from the very first ad impression. It can show us how to find customers more accurately, and how to find better customers.

In short, it can rock your marketing.[Here are a few ways how:

Predictive analytics lets your prospects move through the sales funnel at their pace

As marketers, one of our core jobs is lead nurturing – moving brand new prospects through the various stages of the sales funnel (aka “buyer’s journey”) until they become customers.

We do this by sending well-timed content, by personalizing some of that content, by enticing them to take tiny steps toward our goal. These are often known as “micro conversions” – a white paper downloaded. An online calculator used. A demo scheduled.

Most of you are more than well acquainted with this process. It’s your job, after all.

Well, predictive analytics can let you outsource some of that work. By analyzing tens of thousands (even millions) of prospect actions, it can estimate when each individual prospect might be most likely to complete one of those little micro-conversions.

In other words, it brings people through the sales funnel more effectively than you.

Don’t let that capability make you worry about losing your job – there’s still plenty of work for you to do. But just like it’s not a good use of your time to manually reformat typos in your mailing list ( to, for instance), it’s not a great use of your time to run an assessment on each individual prospect as they move through the sales funnel.

Doing that for just 100 prospects might well take up your entire day. So we let the algorithms of predictive analytics do it. While you go make sure your team members are working well (for instance), and go make sure IT understands the needs of your new app, and…. You get the idea. While you go do the rest of your job.

Predicting demand

By drawing on a blend of data feeds (past sales, current economic conditions, media coverage, social media activity and more) algorithms can be a highly educated guess about which products will be in demand when.

Marketers can use this two ways:

  • Boost what’s already working.

If the system predicts a big upswing in blue fuzzy slippers (for example), marketers can generate content and advertising to push the slippers. They can position their product to be in the right place at the right time to capture demand.

  • Respond to inventory issues.

If there’s predicted demand for the fuzzy slippers, but not enough inventory to cover the orders, the marketer has some options. They could increase the price of those slippers, thus making a higher margin on the inventory they do have. Or they could give their best customers the opportunity to buy those fuzzy slippers first.

That’s just two ways that knowing about demand in advance could help marketers do their jobs better. Think of it as a bit like a weather report for demand generation.

Find similar buyers

If you’ve ever done any advertising, you know about the idea of lookalike audiences. These are individuals who are not currently customers or prospects, but they match many of the characteristics of your best customers.

By picking the right attributes of your best customers, you can ask an advertising platform (like Facebook, for example) to go find people who fit your parameters, and then show your ads to them.

The trick is picking the right parameters. Basic demographics may not be enough to define an audience that’ll go mad for your ads.

That’s where the predictive analytics comes in. By being able to analyze hundreds, even thousands of attributes about your best customers, the predictive analytics system can create a profile more detailed than anything you, the human, would have time to define.

And so the algorithm gets to pick which lookalike audience to advertise to. It may also be tasked with creating the ads you’ll show this AI-selected group. And it might even personalize those ads for you.

Remember – once the predictive analytics algorithm knows how to pick the audience, make the ads, and personalize them, it can zoom through that work at computer speed. The same speed it processes any other data. That’s way faster than the click… type… click… double-click pace we humans work at.

Want proof of how well this works? One Harley Davidson dealership increased its leads by 2,930% in three months thanks to predictive analytics. Half of those leads came from lookalike audiences that the dealership had never before considered reaching out to. But the AI knew just where to find them.

Offering the nicest price

Some of us aren’t going to like this tactic very much. It has a whiff of being sneaky, even kinda sleazy. But alas – it works.

Because the algorithms know so much about us and how we respond to ads and offers and products we’re searching for, they know that we respond to different price points. And so they can offer different prices to different customers.

If that strikes you as unfair, I get it. Some of us are a little cool on this approach too. But marketers have actually been doing this for at least a decade; they were just doing it at a more simplistic level. Catalog companies used to print different prices for people in different zip codes. More recently, airlines and travel websites have perfected the technique.

Here’s how it works: If you live in an ultra-high-income zip code, the price for a particular Christmas wreath might be $175. If you live in a lower income zip code, the wreath would be $125.

Of course, this cuts into the margins the company makes. But if they’re still doing well enough at even the lower price, it’s a win. They’re also getting the benefit of making a sale. Once you’re a customer, they can market to you more accurately and successfully.

For many companies, even if they lose a little bit on the first order, they’ve got a sophisticated enough marketing system to make up the loss later when you buy again.

Create far more refined customer personas

This tactic is similar to segmenting, except it’s more like segmenting 10.0. You’ll be segmenting your customers and prospects based on every data point you’ve got – well, you won’t be doing that, the predictive analytics algorithm will do that.

When human marketers create personas, we tend to have to stick to 3-5 key personas. It’s just too much work and time to create a persona for every tiny little instance. We do our best, of course, but at some point, you have to go home to sleep and you have to address other demands of your job.

So you pick the personas that make up the largest chunk of revenue, you build content and a buyer’s journey that best addresses their needs as you can, and you call it good enough.

And that is pretty darn good. It’s way better than just treating everyone the same, that’s for sure. And this level of segmentation and personas works – you’ll get 50-300% more results just by treating these groups differently.

But compared to what an AI-driven predictive analytics program can do, this is child’s play. The AI can crunch every element of data – terabytes and petabytes of it – to find “clusters” of different persona types. It will see similarities among customers and prospects that humans wouldn’t see unless we had way more time and focus than we do.

The AI can then address those clusters’ needs with content they’ll like most, via channels they prefer, at times when they are most likely to respond.

The result? Dramatically higher numbers of leads, better leads, and leads that move through the sales funnel faster.


We’ve barely scratched the surface of what predictive analytics can do for marketers. This post could easily be extended into a book.

But we have covered enough to show you what’s possible. And hopefully enough to show that predictive analytics aren’t out to steal your job.

Just think of AI and predictive analytics as computers 2.0. They’re a power tool to manage the mountain of data your business accrues every hour.

Predictive analytics and AI are just better tools than spreadsheets and even good CRMs and content management systems. Think of those old systems like a shovel, or maybe even a spade. AI and predictive analytics are more like backhoes and mining equipment.

Marketers get to play with the big toys now.

What's New?