American Marketer


Bring mobile analytics full circle with A/B testing, optimization and retargeting

August 29, 2013

Brendan O'Kane is CEO of OtherLevels


By Brendan O’Kane

Since the earliest days of SMS campaigns, mobile devices’ growing capabilities have brought us increasingly sophisticated marketing efforts.

Yet when it comes to measuring the impact of messaging sent from these devices, 59 percent of marketers still consider themselves “inexperienced” or “novices” at mobile measurement and 53 percent believe there is much untapped opportunity.

It is important to start and continue a discussion of A/B testing, laid out in an easy-to-follow progression, and bring the analytics process full circle with a look at messaging optimization and retargeting.

5 steps to A/B split testing
When designing a set of A/B tests, it can help to think of it as five simple steps:

1. Defining marketing objectives: Some marketers may be looking for increased application traffic, others for a boost in sales and yet others for a higher number of social shares.

2. Segmenting the target audience based on the objectives: For instance, if a marketer wants to increase the number of app use sessions by “light users” segments or the number of social shares by “heavy users.”

3. Creating variations: Based on what marketers know about their target audiences, they can compose messages with varying lengths, tones, wording, calls to action and visual design elements.

4. Deploying the messages: For the sake of example, let us say an electronics retailer is having a back-to-school sale on Apple products. The message split can look like this:

• Message A: Hey Moms! 10% off on all Apple products for school!

• Message B: Back to school special! 10% off on all Apple products!

5. Interpreting the results: For every 100,000 customers who got message A, 30,000 opened it. Of those 30,000 customers, 10,000 clicked the coupon and redeemed the offer. For every 100,000 customers who got message B, 25,000 opened it. Of those 25,000 customers, 12,500 clicked the coupon and redeemed the offer.

In this case message B is the winner. That is because the true winner in an A/B test is the message that drives the highest number of goal conversions, not opens. Half of those who opened message B redeemed the offer, while only 33 percent of those who opened message A did the same.

A/B tests can also be optimized by pre-defining certain measurement criteria such as particular user behavior or outcomes. Marketers send out the message variables, gauge the results in real time and automatically push out the winning message to a whole target audience for maximum efficiency.

However, the story does not end here. What about those customers who opened the message but did not convert and those who never opened the message at all?

Retargeting: Way to maximize ROI from all messages
Back in the 1990s, there were a lot of consumers grumbling about the seemingly never-ending string of “free trial” disks and CDs that landed in their mailboxes courtesy of AOL. One would offer 300 free online hours with a sign-up for the service. The next would offer 500, the next 700 and so on.

Those candy-colored disks and CDs may have set AOL back $300 million, but they also helped the company grow its user base from 200,000 in 1992 to 25 million in 2002. At one point, the company was getting new subscribers at the rate of one every six seconds.

I am quite sure the avalanche of follow-up offers – at one point, 50 percent of the CDs being made worldwide had an AOL logo on them – had much to do with that rapid growth.

Well-informed persistence, as AOL clearly knows, is the essence of retargeting.

When customers do not respond to your messages, you use the information you have gathered about them so far to make follow-up messages more enticing and likely to lead to a conversion.

Continuing with our A/B testing example from above, to retarget, the retailer could:

1. Try sending the same message to those who did not open it the first time around. They may have been too busy at the time or perhaps they just did not notice the message or were not in a “shopping frame of mind.”

2. For those who did open the message but did not redeem the offer, the retailer could add the words “this week only!” to add a sense of urgency or bump up the offered discount to 15 percent to make it more appealing. By sending out several rounds of retargeted messages, the retailer can gain new converts each time.

Of course, there are those who simply are not interested and, for whatever reason, will not ever redeem.

If, after a certain number of follow-ups, a particular customer still has not responded, it is best to remove him or her from the campaign so marketers can focus resources on those more likely to respond.

MOBILE MARKETING budgets are growing, and with them the need to show ROI.

All the variables marketers need to consider carefully can make it a challenge to take the first steps in that direction. But, taken as a series of simple steps as described above, in reality it is easier than it seems.

Adopting a mobile messaging analytics strategy that aligns with your particular business objectives will help you achieve that goal. If you engage customers with great messaging today, you are more likely to keep them tomorrow – and far into the future.

Brendan O’Kane is CEO of OtherLevels, San Francisco. Reach him at