Commentary
Missing the Target with Targeted Ads
- by Chuck Martin , Staff Writer, January 28, 2014
There’s targeting by location and then there’s
missing the location altogether.
At least that’s the impression given from an early morning presentation at Day 2 of the MediaPost Mobile Insider Summit here in Captiva, Florida.
Some somewhat surprising stats regarding location-based advertising were tossed out early in the day and resonated throughout various comments as the day wore on.
The effectiveness of ads near locations, such as quick service restaurants, delivered in a presentation by Eli Portnoy, founder and GM of Thinknear, were not all that surprising.
For example, a study the company conducted found that the number of people who clicked on a mobile ad were much higher the closer they were to the restaurant.
The majority (58%) of those within a half mile of the restaurant responded while fewer (42%) of those between one and one-and-a-half miles responded. Those four and a half miles away could not be persuaded.
However, the more interesting insight from the study was the lack of accuracy in judging location for advertising.
Based on intended location, much mobile advertising was not accurate. For example, the study found the following accuracy rates in ad messages sent to phones:
- 32% accurate within 325 feet
- 43% accurate within 11,000 feet
- 26% accurate within 6 miles
“As an industry, we have to get really, really accurate,” Portnoy told me after his presentation. He said the study data came from tens of millions of ads served and that the data came from ads served based on location at the time.
The company ran ads through a number of ad servers and then contacted recipients to enter their current location, which then were matched to the intended location of the ad being sent.
While the phone itself may have good location data, any given ad being sent to the phone may have, though various routes and location sources, may be somewhat off in presuming the location.
From a commerce standpoint, this could mean an intended ad for a shopper based on their store location could miss the mark. Even by up to six miles.
Chuck Martin is Editor of the AI & IoT Daily at MediaPost and writes the daily Connected Thinking column. He is a NY Times Business bestselling author and his latest book is “Digital Transformation 3.0” (The New Business-to-Consumer Connections of The Internet of Things.) He is CEO of Net Future Institute and a frequent keynote speaker internationally. Follow him on Twitter @chuckmartin and contact him at Chuck@MediaPost.com.
Let me share a personal story that illustrates the obvious downside of all this targeting foolishness: I've listed a pair of Bose speakers for sale on Craig's List. In the process of crafting my ad, I went to several sites looking for comparable prices. The targeting technologies completely misinterpreted my actions as someone looking to buy something. As a result I'm now being stalked by ads for the very speakers I'm trying to sell.
Yes, Mike, that is one of the downsides on online searching/targeting. In this case, the study was only based on mobile location targeting.
Thanks for the stats. The accuracy is higher than I would have assumed. One thing possibly worth noting is that the (low) location accuracy rate is probably priced into the CPM rates at which the ad inventory is being sold -- not necessarily deliberately, but by simple market forces. In other words, in order to sell impressions with low location accuracy (i.e., low performance), publishers must lower their CPM rates to allow for the "extra" impressions advertisers are forced buy in order to reach their campaign goals. At a low enough CPM, I as an advertiser don't really care if the location data is only 25% accurate -- there's no downside to taking a "brute force" approach and running 4x as many impressions. 3/4 of the imps will be "wasted" and appear outside my targeting area, but that's irrelevant as long as my overall campaign ROI is satisfactory.
Thanks, Myles, for adding another perspective to the same data. Appreciated.
I worked on the launch of the out-of-home industry currency audience measurement system here in Australia. As part of that we had to VERY ACCURATELY determine the precise location of 66,000 advertising faces in our major mainland capital cities, in order to determine the visibility of the sign from a passing vehicle. For example, the visibility of a roadside sign is highly correlated to its offset from the kerbside (among other factors). 1 metre further back from the kerbside will reduce its visibility. When we 'pinned' the data onto Google Maps we found some glaring errors with the geo-location data (we took two measures each time for concordance). For example, some of the 'airport drive' signs as you enter Brisbane airport were indicated to be on the actual runway (fine if your target audience is airline pilots I suppose). Down in Sydney we ended up with around a dozen signs in Botany Bay itself. It took us quite some time to find a reliable device which we then synched to laptop software (there were no iPads back then) which showed the data just recorded on a map. We then reverse-engineered it so we could zoom the map and drag the pin back to EXACTLY where it should be (within a half a metre).
Great insight, John, and thanks for sharing such an experience. These solutions are far more difficult than initially thought, as you demonstrate.
Thanks for bringing more attention to this topic Chuck. This is an important topic that advertisers need to understand better. As someone close to this, I am surprised when people are not asking better questions on source for these location targeted impressions. At scale, this is simply not available through exchanges in quality. The exchanges are faking impressions as quality and passing them off as true lat/long. A very small percent of quality location data is available through exchanges. Yet, here's the thing, most mobile networks exclusively work through exchanges and claim to be location targeting experts. As someone who educates clients every day on this topic and the differences between us and these exchange players, I find it amusing when one of them says we have to get more accurate. We already are.
You are most welcome, Michael and great point you raise about the lack of the level of questions being asked.
Really interesting article and thanks for the added perspective Myles. Let hope they do charge lower but from what Michael says looks like they are getting away with it.
Thank you Humphrey. One of the attendees at the Mobile Insider Summit, following the presentation of this data, suggested this may be the elephant in the room.
Amazing to see how the interest in location has exploded over the past year. We love to see the varying views and discussion topics. There is definitely a lot of location noise coming through the exchanges, much of which is of poor quality. Many networks and legacy players simply weren’t built to handle and understand the data. The industry as a whole is moving to programmatic RTB models, which spells huge opportunities for brands and agencies to connect with consumers by using location in a (very) scalable way. Scale isn’t the issue any more. The ability to deliver accuracy at scale matters most. The good news is that agencies and brands have a number of options today. Finding the right partners who deliver consistently and provide measurable ROI for campaigns is increasingly vital.
Thanks, Eli, totally agree about the growth of location proliferation (and importance).