Think twice when we deploy algorithm-based marketing

While we are excited to follow the emerging big data trend for providing smarter, more relevant and personalized content to our customers, we must be careful to decide when, where and what are we showing them. From a recent HBR article, we all can learn a lesson from Facebook’s bad example: their “Year in Review” app showed a father his dead daughter’s photo last year and being bombard by public on their “thoughtless” design.

What happened to this tech giant suggested that our marketing algorithms may still not be sensitive enough to context. To let our digitized engagement activities to be more “humane”, a wider variety of information that matters (affects how the target audiences perceive the content) will be needed. In the case of Facebook “Year in Review” app, the algorithm might need to include semantic analysis so that the system may know to avoid posting a sad memory of a person. We would like to pay special attention to occasions that might easily trigger people’s negative emotional responses, as reminding people of their unhappy experience not only adds no value to the business, but also ruin the relationship between the brand and customers. To start with, you may proactively check if customer would like to update their personal profile on your database annually (e.g. if one of them get divorced, you will not want your system to send them wedding anniversary promotions).

Besides educating our systems to be more “considerate” on selecting content, we shall also need controlling the level of smart interaction to prevent creeping someone out. Privacy is always customers’ concern. It is important to make our customer trust us when they know what we know about them and be well informed on how we are going to use these data. We might be collecting data from various channels while planning to make good use of these efforts in the marketing strategies, but we shall also beware of “surprising” our customers with an unexpected personalized experience.

Here are a few useful tactics suggested by the article that can help:

  • Field experimentation and observational research:
    Testing the impact of different factors and generate more flexible algorithm
  • Introduce unpredictability:
    Avoiding customers being habituated to regular marketing activities by adding excitements / surprises
  • Identify key customers decision and experience points and encourage human interaction:
    Combining the power of technology and flexibility of human to discover new needs or handle exceptional situations

Are you collecting data now for your next smart campaign? It can be very interesting to embrace algorithms in our marketing strategies and it worth for a cautious planning to start. Make sure you consider your customer’s point of view.