Tag Archives: mobile

Improving mobile advertising via colocation

Every spring for the past five years, analyst Mary Meeker has published her influential report on “Internet Trends” and noticed the same, strange thing – while most media platforms attract a share of advertising budgets closely matching the time consumers spend on them, mobile always falls behind. TV, for example, accounts for 38% of U.S. consumer daily “media time” and also attracts 38% of all U.S. ad spend. Radio has 9% of consumer media attention and gets, yes, 9% of media budgets. But for some reason, mobile falls short.

Today, there are 2.8 billion smartphones on the planet – so nearly 1 in 2 humans use a mobile screen – and in the United States, adults spend 3.1 hours per day on mobile, about 28% of daily “media time.” But mobile attracts only 21% of U.S. ad dollars.

So what’s up? Why are marketers still slow to spend on mobile?

The core reason is media plan budget allocations always try to predict, and then adjust to, actual returns in performance. If the key metric return on ad spend (ROAS) shows that one type of media returns $3 for every $1 spent on advertising (an ROAS of 3:1) and another returns only 90 cents (ROAS 0.9:1), then the first media channel will get more budget. Because ad dollars are not flowing to mobile user eyeballs, something in mobile performance must be broken.

Solution: Moving beyond geo-targeting

The American Marketing Association recently offered a solution, suggesting marketers typically get mobile audience data wrong. In the AMA’s most recent Journal of Marketing, researchers Peter Pal Zubcsek, Zsolt Katona and Miklos Sarvary point out that mobile devices are collecting reams of data about consumers, but marketers typically respond with simplistic geo-targeting that does not take advantage of all options.

First, consider the data. Some audience profile data is “sandboxed” or not shared between apps, while other platforms (especially Facebook and Instagram) can target mobile audiences based on hundreds of profile or interest habits (everything Facebook knows about you can be used to reach you on your iPhone). But the most valuable data from mobile comes from user locations – the geographic coordinates that not only reveal where a mobile user is today, but where he or she has gone over time, and consistencies or variances in those travel patterns. People are just smart animals, after all, and our trails give away our true desires.

Alex Pentland, an MIT researcher who experiments with reams of mobile data, was able to predict which movie theaters in winter have crowds of people who are contagious with the flu virus by assessing slight common variances in mobile travel and communication patterns that indicate who is starting not to feel well – in essence, predicting people will get sick before they know it themselves.

Zubcsek, Katona and Sarvary build on this “spacial-movement-pattern recognition” theory, suggesting marketers need to move past simple geo-targeting to consider the social ties between people visiting locations.

Looking outside the fence

Much geotargeting is still rudimentary, focused on audience member X being at location Y so getting offer Z. For example, a current popular mobile targeting approach is to build a virtual fence around an auto dealer, and then hit a consumer with a mobile offer when she or he enters that perimeter. Another, slightly better approach, can pick up people who have visited a certain type of retail location and then retarget them elsewhere later – say, a child clothing retailer could retarget consumers at home who have recently visited daycares, baby-supply stores and toy stores, knowing those audiences are likely new moms. The thesis here is a consumer makes a “location choice” to maximizing some underlying utility function (to buy something, eat something, make a transaction), so marketers who sell similar services simply match their offers to that location signal.

But what if you consider deeper uses of mobile data?

1. Push consumers slightly off path. Shoppers who travel off planned paths tend to be more profitable. Zubcsek, Katona and Sarvary point out a coupon that hits someone on a pre-planned path may have limited influence, or worse, give away too much, for someone who is already thinking of making a purchase at that location. But if a mobile offer pushes that consumer to go slightly further off path, while response rates may decline, the consumer may make a much more profitable transaction. The “effort” involved in that consumer shifting travel patterns means they may have greater intent to respond. There is obviously an inflection point where the economics break down – someone seeking a restaurant on a Friday night in Chicago is not going to pop over to Los Angeles – but if marketers model distance paths and optimal pushes for changing the consumer pathway within an acceptable distance, the resulting sale and profit will be higher.

2. Consider mobile colocation data for modeling. Zubcsek, Katona and Sarvary write that consumers who visit the same pattern of places have similar preferences, even if they don’t know each other – and that by mining location data, marketers can build segmentation models that are 19% more likely to predict profitable purchases vs. older standard segmentation tools (e.g. psychographic and demographic segmentation or media usage studies). In other words, mobile geo data isn’t just for targeting; it can model groups most likely to be your Most Valuable Customers.

3. Go long. When offering mobile incentives, extend the length of the offer and the frequency of impressions. Once targeted appropriately, consumers are more likely to take action on an offer if their future repeat paths put them in proximity to the desired action. This is intuitive, but many marketers miss the mark by not allowing a frequent-enough series of exposures to align ad impression with audience target as he or she nears the desired location of action.

4. Look elsewhere. Beware that ignoring valuable prospects outside a specific geo-fence “may leave money on the table.” Consumers signal their interests when they share similar travel patterns. So if Consumer B closely matches Consumer A in geo-location travel, but only Consumer A moves within a geo-target fence to “trigger” an offer, Consumer B may be just as – or more – valuable! So targeting of consumers via mobile must move beyond specific single geo-fences to a range of geo-triggers, if you are to reach similar prospects with high interest in purchasing your product.

5. Mix up the formats. Creative design obviously has influence in any advertising, and mobile creative options are now much more vast than small banners. In-image ads, video ads, sponsored stories, native ad units, and 360-degree immersion units support higher frequency against targets, once identified, without triggering wearout. A mix of ad units can also boost response; in the advertising research book “What Sticks,” authors Rex Briggs and Greg Stuart uncovered that consumers reached with different types of impressions typically respond at higher levels vs. one media format used at the same frequency. So mix it up.

All of this is to say that geo-targeting by itself is not enough. Marketers’ main goal is to match messages with people most inclined to take action. What Zubcsek, Katona and Sarvary have uncovered in their research is that mining the travel connections of consumers across time and space, and matching those paths, unlocks new ways to identify Most Valuable Prospects likely to respond. By mining these pathways through colocation, you can predict not only where consumers will go, but also what they may do next.

Or, as MIT’s Alex Pentland might say, don’t forget to get a flu shot.