Predictive Analytics in Marketing: Does It Really Predict?

7 min.

Known since the early 2000s, the concept of predictive analytics has only recently turned into a viable technology for identifying the probability of future outcomes based on historical data. Now, this technology is rapidly gaining momentum: the global predictive analytics market is expected to hit $14.95 billion by 2023 at a 21.2% CAGR.

Predictive analytics proves beneficial for almost any industry, but there are a few for which it has become a game changer, including financial services, workforce management, healthcare, manufacturing, and public sector, with marketing topping the list.

In this article, we will look into what predictive analytics has to offer for marketing, outline some helpful tools and try to answer a paramount question: does predictive analytics really predict?

Now, let’s start with the basics.

2 pillars of predictive analytics

Predictive analytics became possible with the rise of two powerful technologies—big data and machine learning. Let’s see how they work together to enable predictive analytics.

Big data

Predictive marketing thrives on data. Our digital footprints, be it an email to a friend or our clicks on a web page, grow into inconceivable exabytes called big data.

With all these dimensions though, data may often be incomplete, flawed, stored in diverse types, formats and locations, which all impairs predictive analytics efforts. According to IBM’s estimate, decisions made with poor-quality data cost the US economy about $3.1 trillion in 2016. Harnessing such data requires a powerful data management solution in place. Another technology for getting to grips with big data is machine learning.

Machine learning

Machine learning is a smart technology that turns bytes into insights. It implies a variety of algorithms to build predictive models that are trained, updated, and retrained regularly to generate clear actionable outcomes. Obsolete models also impair predicting efforts, so regular model retraining is a must to keep and improve the efficiency of predictive marketing analytics.

High-quality data and well-trained relevant models give great results when combined, enabling companies to go as far as, say, Amazon—the only company with a patent that allows them to ship goods before an order is even placed. Most companies, however, use predictive analytics to yield less impressive, yet vivid results. Let’s look into these use cases.

5 use cases of predictive analytics in marketing

With clean data and the right tools to collate it, predictive analytics proves instrumental to business strategies. The companies apply it to marketing for the following purposes:

1.  To predict customer behavior based on advanced customer segmentation and buying propensity, as well as to recommend products and services aiming at targeted upselling, cross-selling, and next-selling.

This worked well with Tesco, a leading UK grocery chain that used customer behavior data to see which products their loyalty program members were buying together. As a result, their shoppers who bought diapers for the first time were also mailed discount coupons for beer because data analysis revealed that new fathers spent less time at pubs and were likely to buy more beer at the store.

2.  To qualify and prioritize leads based on predictive scoring, identification models, and automated segmentation.

After adopting predictive analytics, Shoretel, a Californian telecommunications company, started prioritizing their best-fit leads. As a result, Shoretel reduced the number of cold calls from 100 per 1 most qualified lead (MQL) to 12 per 1 MQL.

3.  To bring the right product or service to market.

Netflix’s House of Cards is a good example of a predictive analytics success story. By analyzing the data from their 27 million US subscribers, Netflix made key decisions that led the series to the Emmy award in 2013.

4.  To personalize content.

Using predictive analytics to deliver the most relevant messages to the right players,, one of the largest lottery intermediaries in Europe, improved their targeting accuracy by 300%.

5.  To determine the most effective marketing channels.

Insights derived from predictive efforts enabled Home Depot, the largest home improvement retailer in the United States, to increase their ROI on mobile ad spend by 800% using location extensions in ads to reach customers who were close to their stores.

It’s clear now that predictive analytics may work wonders for marketing. But how to set it up? Here are some options when it comes to the technology underpinning it.

Predictive analytics for marketing: The options and tools

Setting up predictive marketing analytics requires specialized software. Here, businesses have a few options to choose from: to go for an off-the-shelf product as it is, or to commission a vendor with customizing it or even with developing a wholly custom tool for predictive analytics.

Off-the-shelf products

The tools, including full-fledged business intelligence solutions with built-in predictive analytics, are numerous. Here’s a short overview of the ones scoring high in 2018.


EverString is a SaaS solution for B2B marketing professionals. This tool provides valuable insights into lead scoring, one of the key elements of marketing success.

Relying on predictive models that are based on historical data about converted leads, the tool singles out prospects whose web activity patterns (the so-called signals) are similar to that of the converted leads. With this, the tool builds a list of potential customers. EverString can even sort out potential competitors for launching a marketing campaign to lure their prospects as well.

The tool offers friendly pricing models ($9 - $300 per month) depending on the required functionality.


Alteryx is a powerful analytical platform with a user-friendly intuitive interface that requires a minimal learning curve. The tool can be easily used across many teams, from an IT department to marketing and sales.

Alteryx offers about 50 prepackaged tools that cover common predictive analytics functions, including grouping and forecasting. These tools are drag-and-drop, which spares the need for any programming knowledge.

Besides, Alteryx allows blending data from multiple sources such as Microsoft Excel and Salesforce. The platform also supports a wide range of tasks, from data cleansing to general data manipulation.

Unfortunately, while being rather pricey (from $5,000 to $58,000 per annual subscription), the platform lacks front-end visualization and requires exporting to other sources (Microsoft Excel, Tableau, QlikView and the like) to view the content.


Tableau has become one of the most widely used BI platforms due to its non-tech user friendliness and flexible deployment options (cloud or on-premises). Tableau supports advanced UI-integrated analytical functions, including trending and forecasting.

As for its cons, the users name pricing ($12 - $70 user/month depending on the product) and limited capabilities when it comes to integrating data from multiple sources.

Custom tools: A viable alternative

In case your business has specific workflows, you can turn to a software development vendor to get a custom tool or customize a third-party one in line with your particular requirements. This helps to set up the tool that will fit naturally into your infrastructure with efficient data interoperability between your internal systems and channels.

Summing up, a business of any scale has a variety of predictive analytics solutions to choose from in order to cut the guesswork and make informed decisions relying on historical data. Still, even with a suitable predictive analytics tool, sometimes predictions can’t be made that easily.

Some limitations to keep in mind

When it comes to consumer behavior, the human factor is very hard to predict as it is not governed merely by math. That said, the first and foremost limitation of predictive marketing analytics is that it doesn’t predict an event itself but just its probability.

To make sure the probability estimation isn’t hindered, it’s necessary to address further potential issues preemptively:

  • Poor quality data that may lead to false prediction
  • The lack of skilled professionals, including both competent data scientists and marketing analysts to prepare the data and make assumptions
  • Outdated predictive models built with data that is not relevant any longer
  • The lack of integration between the analytical tools and other business-critical systems within a company

If these issues are properly addressed, you are good to rely on predictive models in making your business decisions indeed.


Powered by big data and machine learning, predictive analytics is a key business tool that made its way through a number of industries, especially marketing. For example, predictive analytics can enable marketers to reach the most valuable leads through the most efficient channels, among other use cases that bring in higher conversion rates, footfall and loyalty.

To enable predictive analytics, marketers can pick from the following options:

1.   Buying an off-the-shelf tool and customizing it

2.   Having a custom-made tool tailored specifically to their needs

Whatever option marketers choose, they should bear in mind the key limitation of predictive analytics: it predicts the probability, not the outcome. So predictive analytics makes a good strategy to get an appropriate set of probabilities to act upon, yet it won’t provide a clear-cut action plan to follow blindly.