The 7 Best Uses for Predictive Analytics in Multichannel Marketing
Caesers has refined its customer datato this single view and thus is able to “determine an ongoing level of re-investment for each player,” according to Zabin. “The predictive score is combined with each player’s profile and offers are delivered across multiple channels—direct mail, email, mobile.” When a customer responds, Caesars updates and recalculates that profile. Customer behavior allows the casino giant to determine which offers—restaurant, hotel, slots or a show—will appeal to which customer and how often to send them, he adds. That way, communications can match the fact that a particular customer only travels to Vegas once every couple years.
Kobielus says once marketers build the single customer view, you should use it on all customer-facing applications, including marketing, sales and customer service.
2. Determine promotional effectiveness not only by channel, but also by narrowly defined customer segments, Zabin says. “Different promotional tactics (e.g., coupons, discounted prices, sampling, special displays, feature ads, on-pack stickers, special packaging, events, etc.) tend to elicit different response rates bases on the characteristics of the target segment. After all, different customer segments have different price sensitivities. Sending a 15 percent discount coupon to a consumer who only buys that company’s product loses money, since that consumer is unlikely to buy a competing product. Sending a 15 percent discount to a brand-switcher is a whole different story.”
3. Notice which customers are already maintaining a relationship with a marketer in more than one channel. Sadh notes that when marketers maintain multichannel contacts with the same customer, it can increase campaign effectiveness because “multichannel users tend to be more loyal,” according to her NCDM 2010 presentation: “Leverage Analytics and Campaign Strategies to Deploy a High ROI CRM and Database Marketing Practice.”
4. Let Social media data inform your multichannel strategies and be part of predictive models. Kobielus says: “Incorporate social network analysis variables into these predictive models in order to assess how sensitive acceptance rates are to hidden, latent, non-obvious connections among people, groups and organizations. … Integrate these predictive models with complex event-processing middleware in order [to] tie targeted offers to real-time variables such as customer geospatial coordinates, portal clickstreams and sentiment analysis on social media messages.”