The topic of artificial intelligence [AI] for Marketing covers many different areas including content delivery, content generation, social media, personalizing emails, attribution, and optimizing lead gen, just to name a few. This blog focuses on applying AI in the marketing funnel for optimizing lead gen and marketing attribution[MTA].
Marketing attribution is near and dear to my heart; I started my career as an accountant, and later switched to marketing. I got into marketing based on questions I heard from a number of CEO’s who said:
“The challenge I have with marketing is I know half of my marketing money is wasted, the problem is I don’t know which half”
The Good and the Bad
Marketing attribution should provide insights to answer this question and help guide activities for optimizing lead gen. Gartner’s report “Top 10 Strategic Technology Trends for 2018: AI Foundation” (March 2018) found that “One in four [Gartner clients] said they will integrate AI into digital marketing”. The conclusion I draw is it has momentum.
The challenge is Gartner’s “Hype Cycle for Digital Marketing and Advertising, 2018” saw “Multitouch Attribution” slide further down into the “Trough of Disillusionment” from 2017 to 2018. * This means the end vision for MTA has a way to go to deliver on the promise. Multitouch attribution (MTA) refers to software and services that help marketers evaluate how much credit to assign to each element of a multichannel marketing campaign. *
The good news is there are still some basic techniques to provide some answers to the question “which of my marketing dollars are wasted”. The bad news is MTA data is noisy, dirty and time significant.
- Noisy because there is a lot of interactions, with many different intentions and data sources. Some are substantive, others are just noise. The same activity is subject to different interpretation depending on the marketing initiative, how that interaction was generated and nature of the interaction. Some activities nurture leads, and others are designed to move leads along through the sales funnel. Context is important.
- Dirty because of missing or incomplete information. It may be hard to identify who created the interaction or what company or account the lead belongs too. Identifying who created the interaction is critical but tricky. The problem is matching the interaction to a known prospect and opportunity. The same prospect often use different email addresses and companies have multiple email domains that all need to be aligned to a correctly connect a lead to a company/opportunity. Most CRM solutions require a manual process to connect leads to companies and their interaction history is disconnected from the company/opportunity in the process.
- Time significant because some interactions are appropriate at the beginning of a buyer/seller journey but that same interaction toward the end of the buyer/seller journey can mean something totally different.
The first use case of AI for the marketing funnel is algorithms and methods to clean up the noise and dirtiness of the data, add context where possible, and sequence the interactions across time at both the lead and company level. Methods can range from simple filters for excluding incomplete information, to algorithms providing context or to match multiple email addresses and sources of information to specific leads, companies and opportunities. This creates a foundation of information to support the other use cases.
The second use case of AI is to classify interactions as significant for the current buyer seller journey across time. On the image below, the small green circles represent seller/buyer interactions along the buyer/seller journey. The “X” axis of the graph represents time, the “Y” axis of the graph is the the AI generated opportunity health score.
A key measure for marketing is lead source. The small circles on the far left are just noise and would skew the lead source measure. The first substantive lead source is probably the first circle on the X axis to the right of April. AI can quantitatively identify that as the lead source, but visualizing it on the graph below and hovering over the small circles to provide details on each interaction provide additional qualitative confirmation and helps users "buy in" to AI.
The third use case of AI is to assess marketing activity effectiveness from simple lenses of “which marketing campaigns contributed to the most “closed-won” opportunities and which marketing campaigns contributed to no “closed-won” opportunities." The diagram below helps visualize the data through the marketing funnel.
The screen above displays the lead funnel from leads generated in Q1 2018, from the campaign “Innovate Your Sales Process”. Out of 172 leads, 24 were closed-won. Looking down the lead funnel by stage shows one of every 8 leads ended in “closed-won”. The average contract value is $16,450 . In the first box on the upper right part of the screen shows we closed 1 deal in the same period as the campaign . Lifetime to date (box on the right highlighted in blue) we have closed 24 opportunities .
This isn’t as much AI as it is reporting. To understand “where marketing money is wasted” you can filter or stack rank this information, sorting campaigns and activities based on the number of closed-won deals influenced. This could identify activities having the highest impact on "closed-won" deals as well as the lowest impact. I've worked with systems in the past that provide sophisticated AI for valuing and marketing interactions , but my experience is consistent with the “Trough of Disillusionment.”
Blog series on “AI for Sales”
This is the third blog in the series.
The first blog was: Artificial Intelligence [AI] For Sales Forecasting
The second blog was: AI for Sales - AI/Machine Learning Primer for Sales
The next blog will be on: AI for Sales - AI to enhance data collection and data quality.
*Gartner, Hype Cycle for Digital Marketing and Advertising, 2018, 25 July 2018