Sure, it may be appealing to return to a site that appears to recognize you – offering you recommendations on new offers or suggesting a somewhat easier path to finding what you want. Maybe there is one less form to fill out before reading that premium article. Or you are offered another product similar to the one you purchased last week. When these offers are relevant, they are a great way to build a relationship between the company and the visitor.
However, when the offers or suggestions miss their target, do companies really benefit from this personalization? How many times will United Airlines offer me its latest “special” to the Caribbean before it realizes that I absolutely hate hot weather and have booked my past five trips to ski destinations? The more United presents offers that have little to do with my personal past history, the less likely I become to view the airline’s suggestions.
Does this level of personalization improve your relationship with a business – or is it just a reminder that the site is systemically tracking every little thing you do? Unless custom elements are relevant – and designed to further you down the path toward a specific goal – this type of broad-based behavioral targeting may end up creating more of an annoyance and lack of connection than anything else.
Steps to Success
Using behavioral targeting programs to extend or strengthen a relationship with an online customer is really no different than traditional relationship building. Start out by understanding who the prospect is, learn a bit more about how they make decisions (are they emotional buyers or detailed, research-oriented thinkers), and then understand what type of things they are interested in. Making suggestions or recommendations too early on (without enough background information) is akin to the waiter suggesting the gentleman would like the steak – without knowing the guest is a vegetarian.
The same holds true online. When establishing a relationship-driven behavioral program, consider the following:
Establish Clear Goals From the Start
Are you trying to increase order size through cross-selling, are you trying to get customers to buy a more expensive product than the one they’re looking at, or are you trying to get them to place more orders more often (or some combination of all of the above)? Depending on your goals, you may need to collect additional data to create your behavioral program.
Set Clear Rules and Behavioral Triggers
Ensure you have a solid base of data from which to draw up initial targeting rules and identify the correct variable for customer profiles that indicate the desired behaviors. Focus on multiple purchase history or multi-site visit data – as they tend to typically be better predictors than any other variables. If you don’t have extensive depth of data, consider broadening the data points needed to trigger an offer.
Test – Then Optimize
Once you set it up – test, test, test. Having the hypothesis is the first step – executing on it, even better. But an amazingly high number of companies don’t take the final step and measure the performance. Seems like common sense, but sometimes people must be reminded to review the data! Make changes to optimize your programs based on your test results. You may be surprised at how certain sub-profiles or surprise trends emerge several weeks into your program.
Finally, automation is the key to scaling these programs. For companies with a few products, you can set up and optimize the rules manually. As your product mix grows and as the complexity of the targeting schema increases, an automated platform is necessary.
Following its acquisition of the rights to show Champions League football, BT Sport has been working to establish itself as the major rival ... read more
We talk a lot about content. How to make it, what makes it work, how to measure it’s effects, if there’s too ... read more
Sport England wanted to encourage women to increase their physical activity, so it created the campaign ‘This Girl Can’ and its authenticity ... read more
You'd never choose a car based just on a name, so why choose audiences that way? Making targeting decisions based on segment names degrades the data's quality.