• by Kevin Marcus , Op-Ed Contributor, October 15, 2018

Marketers may not know it yet, but we are facing an AI tsunami -- and it’s time to prepare for it.

Artificial intelligence and machine learning are rapidly eclipsing traditional targeting and segmentation methods, and for good reasons. Machines are powerful. AI can analyze thousands of magnitudes more data, far more quickly and accurately than even a team of humans.

AI is smart. It routinely uncovers complex data patterns that humans miss. And most importantly, AI and machine learning  models can be customized to solve specific marketing and sales problems far more quickly and cost effectively than humans.

What we are all witnessing is the sunset of the golden age of traditional targeting and segmentation, due to the sheer power, efficiencies, cost savings and ROI potential of AI. Underlying market pressures are coming together to fundamentally change how we will acquire, serve and engage customers.

For example, how will you respond when a direct competitor begins using AI and machine learning models for targeting and experiences the typical three to four times magnitude improvement compared to traditional targeting? As your competitor’s ability to target prospects improves, the likely impact is that your company will begin losing market share. Also consider that this same competitor will be gaining valuable insights from their AI models and applying what they learn to the type of data they collect. This will further improve their targeting. Once a smart, data-driven marketing engine like this is in place, it makes it even harder to compete.

To avoid being swept away by this AI tsunami, there are four steps you can take today.

First, learn about customizing AI models, which can and should be customized for marketing. Customization requires training. Whether your email campaign targets individuals or households, it's important to avoid the pitfall of sending a coupon to everyone in the household. Circumvent this by systematically scoring household individuals to optimize model performance and train your model future campaigns.

Secondly, recognize how your data impacts models. Although models can run with just 10 to 20 attributes, that data set is likely too little to achieve desired outcomes. AI models love data. If you determine the data attributes you need in advance, you can start collecting that data now to build a smart, AI-driven model down the road.

Understand how problems/assumptions are formulated in AI models. Your model can be skewed if you make the wrong assumptions with your data. For example, an academic-industry client was weighing prospective students based on their interest in school. However, after AI was used to analyze the data, the optimal indicator turned out to be those who applied to school, not those that showed an interest.

Finally, understand how the data impacts what you are trying to solve. Analyze the data you currently collect to determine how you can work with it. Most  targeting companies using AI can analyze your data to determine what problems you can solve and fill any data gaps to solve other desired problems. And, if you’re not already tracking converters v. non-converters or other contrasting data, start now. Not only is it a great way to uncover data gaps, the best models require two lists. Contrasting data reveals which prospects to find more of and which to avoid.

AI and machine learning models are here and their impact on marketing and sales is growing. It’s time to set aside traditional notions and preconceptions of what qualifies as a good prospective customer. AI is being used to  reveal hidden truths every day. You may discover the most predictive indicator of a prospective client is something completely different than you imagined. And, it’s far better for you to discover this before your competition does.