“According to QuanticMind, 97% of leaders believe that the future of marketing lies in the ways that digital marketers work alongside machine learning-based tools.” — Digital Marketing Institute

WORKING WITH CRM (B2B) CAMPAIGNS

In B2B marketing, enterprise companies use Customer Relationship Management (CRM) software such as Salesforce, Oracle, and SugarCRM to keep records of contacts in business partners who maintain key roles in decision-making and purchasing. This data can be used in machine learning environments with marketing analytic software to chart the effectiveness of advertising campaigns to specific markets.

• The inclusion of personalized events like outbound calls and email newsletters in the sales channel can be recorded in charts to analyze results in future purchases.
• Attendance at trade shows and promotional events are noted with time-stamps in marketing analytics focusing on sales cycles.

A large challenge in B2B marketing is attributing the decision-making within a sale, since there may be several people involved in a single purchase. The personalization of B2B marketing can scale to one million events per contact in the generation of charts and analytics from programmed variables.

WORKING WITH E-COMMERCE (B2C) CAMPAIGNS

In B2C marketing analytics, there is less difficulty in attributing the decision-making for a sale to a single person. In B2C marketing, there is not the same opportunity to track business organizations and companies with specific purchasing agents. The retail consumer market represented by B2C is much broader and defined by e-commerce platform requirements.

• Online stores track consumer sales channels on websites and mobile apps with various forms of cookies, where the largest enterprise sites frequently support 100 to 250 million registered users on their platforms.
• This leads to the potential of personalization of the e-commerce data from automated marketing analytics driven by machine learning and data science in real-time.
• Businesses track the variables related to search engine traffic, repeat customers, PPC advertising, and direct email campaigns to determine the effectiveness of advertising in sales cycles.
• These variable events can also chart the effectiveness of limited-time sales and holiday promotions through marketing analytic software.

CALCULATING ROMI ANALYTICS FOR CAMPAIGNS

Return On Marketing Investment (ROMI) is an OPEX model that compares non-CAPEX spend to yield. The term was coined by Guy Powell in 2002 to be used in the evaluation of marketing campaigns with the investment in expensive commercial advertising designed to build long-term brand identity, customer loyalty, and goodwill. ROMI is calculated on the simple formula related to change in revenue, margins, and program spend.

• Powell recognized two kinds of ROMI, “Fuzzy” and “Sharp”.
• Fuzzy ROMI attempts to estimate brand loyalty and goodwill through the statistical measurement of qualitative values in customer feedback.
• Sharp ROMI is used in 90% of marketing analytics to track the total program spend vs.new revenue generated in a business cycle.
• Sharp ROMI metrics are used to evaluate the effectiveness of B2B & B2C advertising with machine learning and data science.

“Big data” repositories are filtered by machine learning algorithms to help decision-makers discern patterns through chart analysis to predict similar success. The ideal is to create a live data science model trained to deploy custom content and displays to customers based on their historical likes and preferences. There are benefits to both prescriptive and descriptive models.

WORKING WITH DESCRIPTIVE MODELS

Descriptive models filter “big data” from e-commerce and CRM resources to evaluate ROMI in the rear view mirror, i.e. through an analysis of past activity.

• Marketing analytic software prioritizes recent activity in sales cycles and should be retrained daily in a data science notebook.
• Comparative information includes campaigns, results, and costs as variants of success across sales channels.

WORKING WITH PREDICTIVE MODELS

Predictive models in marketing analytics are based on live feedback and depend on live data. This information from clients and customers must be stored in a solid, secure, and fast database.

• Predictive models of marketing analytics assist decision-makers to discern how to change advertising campaign content to match consumer tastes and trends.
• Predictive models are developed in the ROMI phase to optimize advertising campaigns through both product-side and customer-side recommendation engines.

These methods can help e-commerce platforms and stores avoid shopping cart abandonment by customers. In marketing analytics driven by data science, the more that you refresh the models, the smarter the results get over time.

OPTIMIZING SUCCESS IN MARKETING ANALYTICS

In order to succeed in marketing analytics for advertising campaigns with machine learning, businesses must separate existing customers, new business, and renewals.

• In e-commerce, most platform revenue is earned from repeat business, which can be optimized through the creation of personalized displays to the customer through product recommendations.
• In preparing data for metrics and analytics in data science, businesses need to collect customer event variables into a single stream, then tag each event, and aggregate the information at scale for the platform.

Champagne charts can be used to determine when path behavior emerges from data analysis.

 

 

Pattern fitting shows how people consume product evaluations before they buy a product.

 

 

CONCLUSION

1. Get the data. Be patient; this will take weeks. It takes a lot of time to get all permission to find out where all the systems are and go from there.

• Get access to all systems
• Agree on a single timestamp method
• Agree on what “success” means for a sale (when, what, where, entity)
• Find all mid-point systems
• Agree on an organizing campaign model
• Single stream data per person/organization/sale, product family/sales motion

2. Use a cloud database: Do not store your data in flat files or on a laptop. Plan for scale, even if you’re just kicking the tires. I really like Snowflake because it’s fast and it’s cloud-based. So if you want to transfer a project from Europe to the United States for analysis it’s available to you. If it’s in your data center, it may be 50–100 times slower to get it across the ocean.

• Reorganize and normalize data
• Anonymize it on input, but leave a decoder system (as people drill into “proof)
• Plan on scale and long-term global inputs/outputs — data is a movable feast
• Enable for future why/what/when queries — add a few helper columns and tables
• Hook it up to a fast, scaleable data science notebook

ADDITIONAL RESOURCES

• This blog is an excerpt from Grover Righter’s webinar, “Using Data Science for Marketing Analytics. Watch the entire webinar here.
• Zepl’s data science platform lets data scientists and analysts rapidly prototype models in Python, R, Scala and SQL to create rich visuals for their marketing analytics use cases. Try it for free at www.zepl.com
• Snowflake helps marketing analytics teams drive more business value from their data by offering a scalable, elastic, and secure cloud data platform for all of their data workloads. Try it for free at www.snowflake.com

ABOUT THE AUTHOR

Grover Righter is the Chief Data Scientist at Zepl. He is focused on science-based marketing programs and runs a backplane of analytic engines measuring email, SEO and social media effectiveness. He has worked on every major know marketing automation platforms including Eloqua, Marketo, Pardot, HubSpot and Leadformix.

Grover has been working in the high technology sector since 1981. He began as a mathematician and design engineer and has been instrumental in the design and development of major technologies at RMS, Inc., AT&T, Unisys and Novell. But every time Grover built a product, he was immediately recruited to ‘sell it’, irrespective of the presence of another sales team within the organization. Eventually, Grover used up his lifetime limit of sales meetings as an engineer and was moved into marketing.