ANALYTICS DRIVEN MARKETING
Companies have been leveraging in-database analytics from Fuzzy Logix to improve marketing effectiveness and stay competitive in today’s global economy. The applications address important areas including campaign management, product recommendations, market basket analysis, churn management and behavioral analysis. These solutions work on a very large volume of data and are embedded in reporting tools commonly accessed by business users. The common driver for the success of all of these applications is the use of in-database analytics.
By leveraging Fuzzy Logix’s in-database analytics to run advanced analytics completely within the data warehouse, customers have eliminated the need to move the data to separate analytic platforms, and as a result have achieved enormous processing efficiencies. In most cases, our customers have achieved throughput that is 10 to 100 times
CUSTOMER BEHAVIOR ANALYSIS FOR TARGETED MARKETING
The amount of data generated by systems that track consumer behavior is huge. Combining purchase history, demographics, and Internet and social media data provides a rich, but enormous amount of data. One of the challenges is determining the demographic characteristics of customers and prospects. By collecting demographic information from a small sample of customers through tools like loyalty programs and surveys we can use statistical techniques to project the viewing and surfing behavior of the known sample and find the nearest neighbors of those customers in the total population. We can then assign the most likely demographic features to each customer based on the fact that people who have similar Internet, social media and buying behavior will have similar demographic characteristics.
Since advertisers are constantly working to target the right audience, knowing the demographic characteristics of viewers will provide additional targeting guidance. Positioning advertisements in the appropriate slot so that the appropriate audience (age, income, urban/rural, etc.) is targeted for the underlying product or service will yield improved results.
A related approach to targeting is behavioral segmentation. Our programs use pattern recognition to cluster segments based on the behavior of buyers.
Traditional segmentation methods involve picking some number of categories (age, income, etc.) and running regression analysis to test the validity of the suggested category. We do the opposite. Our algorithms review your data and create segments based on customer behavioral patterns. Once the segments are identified we calculate the key characteristics that drive differing behavior. The results are free of presupposition and represent true distinction and behavior amongst customer groups.
PREDICTIVE CAMPAIGN MANAGEMENT – IMPROVING CUSTOMER ACQUISITION AND REDUCING THE COST OF ACQUISITION
Enterprises face stiff competition and need to continuously campaign in order to increase or maintain their market share. Marketing executives have to effectively manage these campaigns so that cost of acquiring customers is optimized and the campaigns yield the desired results. When done correctly, the benefits are powerful.
One of the best ways to manage campaigns is to use predictive analytics and target those who have a higher than random chance of responding to a campaign. This type of predictive modeling can be performed using a combination of techniques. Binary choice modeling, where the outcome is either 1 – positive response or 0 – no response can be used understand how several hundred demographic factors, such as age, income, household size, home value, household location, etc. can be used to predict the likelihood of response. Marketers can then use these predictive models to isolate the customers that have a probability of responding higher than a threshold; say 5% or 10%. Using this type of approach will optimize the cost of acquisition and ensure that a desired revenue outcome can be achieved in a given period.
Trident Marketing presents a great example of the benefits of using predictive campaign management. In 2007, CEO Steve Baldelli made what he calls a “revolutionary” change in how he approached his business and grew from a $5 million company to a $53 million company in four years. How did this change happen? Fuzzy Logix and Trident deployed analytics that helped them:
- Optimize marketing and reduce their cost of sale by 50%
- Boost CPC performance which raised the volume of sales by 10%
- Optimize sales processes resulting in a 10% increase in revenue per call
- Increase revenue & profitability by 1,000% in 4 years
Two case studies that detail the successes of Trident are available on our Pervasive Analytics page.
RECOMMENDATION ENGINE – CROSS-SELLING AND UP-SELLING TO CUSTOMERS BY RECOMMENDING APPROPRIATE PRODUCTS
A common and proven tool for cross-selling and up-selling in the media and web-content industry is the recommendation engine. Once the viewer watches content, the provider recommends additional shows based the similarities of viewing behavior and analysis of characteristics like the preferences of similar viewers, genre, parental rating, etc. The better the recommendation engine, the better the chances of viewers subscribing to additional content based on the suggestions of the engine.
Our initial research was based on the data that was publicly made available by Netflix a few years ago. The engine uses statistical methods to infer the movies that should be to recommend to a viewer based on their behavior and the behavior of others like them. With this data, we demonstrated that recommendations can be made instantaneously even when analyzing 20,000 movies and the associated feedback from 2 million viewers. We’ve since improved the model so that it can be used to provide recommendations for many types of products and services and have built recommendation engines for clients in cable, satellite, music and publishing industries.
Once the next likely purchase information is available it can be used to suggest products and services that the customer would not only be interested in learning about, but also which have a high likelihood of being purchased. Companies can use this information with direct mail and email programs and to deliver dynamic offers on the Internet. Sales teams can also use this information to more accurately recommend products and services to their customers and prospects.
PREDICTING CUSTOMER CHURN AND IMPROVING RETENTION
A major concern for media companies, retailers, and cable, telephone and wireless providers is customer churn. It costs quite a lot to acquire a customer and once acquired, it can take months to achieve positive financial impact. If customers choose to defect early, or decide against future purchases, the resultant ‘negative customer lifetime value’ poses major challenge.
The answer to managing customer churn lies in predictive modeling. Using historical information such as payments, interactions and behavior, viewing and usage patterns, quality of service and demographic data, we can predict the likelihood (and reason) a specific customer will defect in near term.
With this information, companies can pre-emptively intervene and offer appropriate incentives for retaining their customer. Because offering retention incentives has financial impact, they cannot be offered to all customers. Offering these incentives based on predictive modeling serves two purposes – reducing potential churn and optimizing the cost of retention.
For one project, we reduced churn by 10% resulting in a material uplift in annual revenue. For a large cable company we analyzed over 400 variables to understand the key drivers of churn and predicted that we can reduce churn by 6.9%. Both projects were completed in less than 8 weeks.
Fuzzy Logix has extensive experience in these types of applications. Our clients have recognized material improvements in response rates and costs resulting in more effective use of marketing time, money and talent.
Please contact us to discuss how our solutions improve business performance.