Predictive Marketing Blog

AI-Assisted Consumer Targeting 101 – Part 3: Profiling, Scoring and Takeaways

AI was just named the Marketing Word of the Year 2017. It’s kind of a big deal. As we’ve said before, AI is one of the hottest growth areas in marketing and modern marketers need to learn to harness the power of predictive analytics and AI to stay competitive.

Data-backed insights will be integral to developing winning marketing strategies in the new year. In order to stay a step ahead, it is more important than ever that marketers effectively implement predictive AI to maximize campaign ROI by targeting their best prospects and most-likely campaign responders.

Part 1 of this series provided an overview of the predictive modeling basics. Part 2 dug deeper into using segmentation, wrangling and modeling to help solve unique business problems. In this last installation, we’ll look at how profiling and scoring can improve campaign response rates.


Profiles are generated during the modeling phase and help paint a picture of who your targets really are. There are two main types of profiles: customer and predictive.

Customer profiles identify demographic, socio-economic and psychographic data based on your uploaded data file. Predictive profiles do much the same, but for your prospects. Each profile gets into the nitty-gritty of what makes your customers unique. Backed by data, these profiles identify shared or linked qualities which lead to interest in your product or service.

With the Reach predictive platform, marketers can upload their customer data and get customer and predictive profiles detailing:

  • Prospect Maps (Look-alike Modeling only)
  • Key Predictors
  • Demographics
  • Buying Activity
  • Finances
  • Donations
  • Lifestyle
  • Assets
  • Neighborhood

Customer and predictive profiles give marketers a more complete understanding or view of just who their customers and prospects are, allowing for highly targeted future campaigns.


Before we jump into scoring, let’s take another look at response modeling. Response modeling analyzes mature campaigns to predict future campaign responders.

Specifically, response modeling can be used to optimize campaigns by identifying your most-likely responders, reducing marketing budget and scoring the people in your customer or prospect file.

So what does scoring do?

Scoring differentiates the population of your current customer file or net-new customers by giving them a score indicating their predicted inclination to respond. The Reach platform scores from most likely to least likely to respond and sorts these responders into deciles knowing that the impact of the campaign response is different for each decile.

Who you decide to include or exclude from a campaign – by choosing to include top responders or suppressing the worst responders – will have a significant impact on how value is added to your campaign.

Prospect Lists and Scoring
Marketers can optimize campaigns by applying models using the scores to select the top two or three responder deciles to get a better response rate on their next campaign. This way, money is not wasted by marketing new campaigns to people who are unlikely to engage.

Using the model scores and lift chart provided in your predictive profile, you can identify the response rates of your scored list by decile and then make a decision on what percentage of people you would like to include in your campaign. Marketers can maximize their campaign ROI by selecting only the top responders.


Hopefully, this series has shown you that consumer profiling isn’t as daunting as it seems. With AI-assisted consumer targeting, marketers can be sure that they increase future campaign performance by using predictive profiling and consumer insights.

Stay in the know by developing a winning predictive strategy with Reach Analytics. Our predictive platform takes the long, drawn-out data cleansing, wrangling, modeling, profiling, and scoring process and gives you access to your best prospects in minutes.

Happy predicting!