Predictive analytics as a tool to increase marketing efficiency

Analytics Strategies

The growing ability to collect and store data has provided businesses with enhanced capabilities for retrospective and real-time analysis. Now we can trace patterns and draw conclusions about failures so as not to step on the same rake. Or we can identify the most successful solutions and repeat our success.

Predictive analytics is always more effective than retrospective or real-time analytics in the long term, just as prevention is more effective than urgent medical care. Retrospective analytics is essentially an autopsy — an analysis of a mistake that can’t be undone. Real-time analytics is an ambulance responding here and now, and predictive analytics is preventive medicine that saves you from the disease in the first place.

bonus for readers

30 handpicked Google Data Studio dashboards for marketers

Download now

The concept of predictive analytics

As Thomas Davenport said, no one has the ability to collect and analyze data from the future. But we have the opportunity to predict the future using data from the past. This is called predictive analytics, and in fact, many companies are already using it. You can use data from the past to:

  1. calculate a customer’s lifetime value (CLV). This indicator will help you understand what value a client will bring your company throughout their lifetime, including future earnings.
  2. develop optimal recommendations based on user behavior data from your website.
  3. predict what products or services a customer is likely to buy in the future.
  4. predict customer churn.
  5. develop a plan and forecast for sales in the next quarter / six months / year.

All of these are simple forms of predictive analytics. Let’s look at popular predictive analytics methods.

Predictive modeling

We can identify the following stages of predictive modeling:

  • Primary data collection
  • Statistical model formation
  • Forecasting
  • Checking / revising the model as additional data becomes available

Predictive models analyze a user’s past behavior to assess the likelihood that they will exhibit certain behavior in the future. This type of analysis also involves models that find subtle patterns in data, such as to detect fraud.

Often, predictive models make calculations immediately when a user passes through the conversion funnel on the way to performing a conversion action — for example, to assess the probability of a user’s achieving a goal. With accurate data on the likelihood of a transition from one step in the funnel to another, a business can better manage the factors that prevent or help users from moving through the funnel and can more accurately describe the patterns of behavior of different categories of customers.

Uncover hidden insights in your marketing data

Where can you use predictive analytics?

The average user has approximately 50 applications on their smartphone. Each of them receives, transmits, and generates data. This data is stored in different services and in different formats. While at first glance this may seem like a positive factor for marketers, working effectively with such a volume of structured and unstructured data is a problem.

Let’s look at a few examples of companies that have successfully applied the results of predictive analytics.

Amazon uses predictive marketing...

… to recommend products and services to users based on their past behavior. According to some reports, such recommendations bring up to 30% of Amazon’s sales. In addition, Amazon had plans to develop a tool that, based on forecasts, would deliver products to zones in which orders were expected even before those orders were placed on the site, reducing the time for delivering goods to customers.


The Macy’s team took advantage of predictive analytics for more accurate direct marketing. Over the course of three months, the company increased its online sales from 8% to 12% by capturing data on product categories browsed by users and sending personalized emails accordingly.

Harley Davidson uses predictive analytics...

… to target potential customers, attract leads, and close deals. They identify the most valuable potential customers who are ready to make a purchase. Then a sales representative contacts these potential customers directly and leads them through the sales process to find the most suitable offer.


StitchFix is another retailer with a unique forecast-based sales model.
When registering with StitchFix, users complete a survey about their style. Then predictive analytics models are applied to offer customers the clothes they’re most likely to want. If customers don’t like the clothes they receive, they can return them with free return shipping.

Sprint uses AI algorithms to identify customers at risk of churning...

… and preventively provide necessary information on how to retain them. Sprint’s AI predicts what customers want and provides them with offers when they’re at the highest risk of leaving the company. Since introducing this AI system, Sprint’s churn rate has plummeted, and customers have given the company excellent ratings for personalized service and targeted offers. As you can see, forecasting customer churn is a feasible task for predictive analytics among SaaS and e-commerce businesses.

Here’s a list of the most popular metrics within the purview of predictive analytics:

  1. Client outflow ratio (churn rate)
  2. Sales plan forecast
  3. Customer lifetime value

How can you implement predictive analytics?

Introducing predictive analytics is impossible without the cooperation of the marketing and analytics departments, understanding the objectives of the study and established order in the data. Performing predictive analytics goes as follows:

  1. Define your hypothesis
  2. Collect data internally and externally to build a model
  3. Define metrics to measure the accuracy of your model
  4. Use a ready-made service or develop your own:
    1. Build an MVP
    2. Train the model in terms of lack of accuracy parameters to achieve a stable working version
    3. Create an interface or report
    4. Update or retrain the model to meet new requirements

At the data collection stage, make sure you’ve set up end-to-end analytics, since without it, implementing predictive analytics is usually ineffective.

Predictive analytics services

The percentage of business decisions based on marketing analytics reached a peak in early 2019 (considering data from the past six years) according to the CMO Survey: Spring 2019 report by Deloitte. According to a study by MarketsandMarkets, the market for predictive analytics will grow from $4 billion to over $12 billion in 2022

An interest in marketing analytics in general — and in predictive analytics in particular — encourages companies to develop easy-to-use solutions and services that make predictive analytics more accessible for businesses.

Here are some of these services:

OWOX BI Insights

  • An OWOX BI product that helps companies achieve marketing goals and grow 22% faster than the market average.
  • The leader in the spring and summer 2019 rankings by G2 Crowd in the categories of “Marketing software – analytics” and “Software for analytics in e-commerce.”
  • Sends forecasts about the implementation of your marketing plan directly to your email.
OWOX BI product


  • Combines marketing data from various sources, making it available for analysis in Google BigQuery.
  • Determines the value of each user’s step using its own funnel-based attribution model.
  • Automatically builds reports to analyze marketing effectiveness.
  • Shows how your sales plan will be implemented, what your growth areas and weaknesses are, and how your market share is changing.

Stay ahead of the curve with smart analytics with OWOX BI

You can learn more about OWOX BI in our article on how to predict growth areas and risks in a marketing plan based on data.


Predictive models offered by Infer will help you combine all your data sources to get a complete picture of the position of your leads in the sales funnel. Infer tracks signals from online sources and public databases, then creates predictive models based on previous main accounts and the rules you set. The data obtained by Infer will be useful to marketers and sales specialists both for finding leads who are likely to convert into customers in the future and for optimizing the sales funnel as a whole.


Radius provides several data analysis services with a focus on predictive B2B marketing. Key features include:

  • Radius Customer Exchange (RCX), which compares your company profile with that of other companies that have the same audience, giving you the opportunity to work together and create your own marketing lists.
  • Radius Connect: Submit predictive data to Salesforce.

The Radius platform also helps marketers exchange data between departments and find new accounts in internal databases. Like Infer, Radius is a cloud-based system.


Based on the rules of predictive modeling, BOARD works in an adaptive interface with real-time dashboards.

This means you can model various scenarios and analyze the possible results without having to create a new model each time.

BOARD comes with several built-in connectors, so you can extract data from almost any source — your ERP system, cloud database, OLAP cube, and even flat files. You can also turn your forecasts into custom applications using BOARD’s tools.

TIBCO Data Science

TIBCO Data Science is a relatively new product, announced in September 2018. Created as a single platform, TIBCO Data Science combines the capabilities of previous generations of services from TIBCO: TIBCO® Statistica, Spotfire Data Science, Spotfire Statistics Services, and TERR.

The Data Science service helps organizations innovate and solve complex problems faster, quickly converting forecasts into optimal solutions.

SAS Advanced Analytics

SAS has a 33% share of the predictive analytics market and 40 years of experience; they provide users with advanced data analysis capabilities based on many visual editors. The main functionality of SAS Advanced Analytics is based on graphs, an automatic process map, embedded code, and automatic time rules.

According to user reviews, SAS Advanced Analytics does an excellent job of predicting and analyzing overall movement and can process large data sets relatively quickly. SAS provides free demos of its products and a knowledge base to help you start working with them.


This software allows you to automate the creation of reports based on time intervals. You can import your own data sets and export them to other programs thanks to more than 60 built-in integrations.

Extensions provide greater flexibility (anomaly detection, word processing, web mining), but may not be included in the basic subscription price

Although RapidMiner was created for data scientists, it’s easy to install and get started with.


IBM SPSS uses data modeling and analytics based on statistics. This software works with structured and unstructured data. It’s available in the cloud, locally, or through a hybrid deployment to meet any security and mobility requirements.

You can use your existing data to build predictive models in the SPSS visual editor and modeling dashboards. Premium support for unstructured data includes linguistic technology and natural language processing, so you can include data from social networks and other text-based sources in your models.


SAP HANA provides databases and applications locally or in the cloud. This software reduces the time required to create models with additional connectors for large external data sets and intuitive visualizations.

You can also connect predictive analytics libraries (PALs) to SAP HANA to get extra insights from large data sets. For client-centric industries, this software offers text and social media data analysis to predict future customer behavior and recommend products based on past behavior.

SAP HANA is compatible with the R programming language, so you don’t need to learn a new language to configure your queries. When your system integrates enough internal data, predictive models automatically provide new insights.


Predictive analytics in marketing is a powerful data science tool whose capabilities can’t be covered in one article. Let us know in the comments which aspects of predictive analytics you would like to learn more about in our next articles.

As a reminder, here are the three commandments of predictive analytics:

  • Start with the basics: check the quality of your data and collect it automatically to eliminate human error. The quality of your trained model depends on the quality of your training data.
  • Never go far from the goal of your research, since it’s not the process that matters, but the result.
  • Observe accuracy requirements. Remember that the results of your forecast can only be validated by measuring how accurate the proven model is when applied to your data.
Say goodbye to data silos with OWOX BI

Useful materials


Expand all Close all
  • What is predictive analytics in marketing?

    - Predictive analytics in marketing refers to the practice of using historical data, statistical algorithms, and machine learning techniques to analyze and predict future customer behavior and outcomes. It helps marketers make data-driven decisions, identify trends, forecast customer preferences, and optimize marketing strategies for better performance.
  • How can predictive analytics benefit marketing strategies?

    - Predictive analytics can benefit marketing strategies in several ways. It can help identify high-value customers and target them with personalized offers, improve customer segmentation and targeting, optimize marketing campaigns by predicting response rates and conversion probabilities, prevent customer churn by identifying at-risk customers, and enhance marketing ROI by allocating resources effectively based on predicted outcomes.
  • What are the key challenges in implementing predictive analytics in marketing?

    - Implementing predictive analytics in marketing can come with its own challenges. Some common challenges include gathering and processing large volumes of data from various sources, ensuring data quality and accuracy, selecting the right predictive models and algorithms, integrating predictive analytics with existing marketing infrastructure, and managing privacy and regulatory concerns related to customer data usage. Effective communication and collaboration between marketing and data science teams are also essential for successful implementation.