Whenever an ad technology comes along promising to attract new advertisers and higher CPMs, sales divisions begin to listen. If it claims to increase the supply of high value inventory, they listen a little bit more. When the technology actually begins to show up increasingly on RFPs it makes the leap from gimmick to serious consideration. Having reached this stage of serious consideration, ad operations is called in to evaluate the product and assess its value.
A lot has been written about behavioral targeting of late, and plenty of it in this publication. Having been involved with at least three companies in the consideration or implementation of the technology there are some interesting challenges as well as benefits from the ad operations standpoint. So hopefully this article will add some new dimension of interest from that perspective-- and will reveal aspects of behavioral targeting not often brought to the surface.
Create a segment, watch what happens
First, so we don't create any international incidents here, let me state what most of us in the business already know: behavioral targeting is based on anonymous information. We only know what groups or "segments" do-- not who they are.
That being said, when you create a segment in behavioral targeting (for example, viewers of SUV product info who have looked at 10 content pages in the last two weeks) there is an interesting side benefit that goes beyond just serving it an ad. You can view where and how frequently the segment traverses other areas of your site, and that gives you a more complete picture of their profile as a user. It might interest you to know that out of the 600,000 pages viewed by people in this SUV segment, 50,000 page views were generated when they looked at content related to recreational road trips, 50,000 in local weather, 50,000 in motorsports, 100,000 in run of site (RON) pages, and the balance generated in pages devoted to SUV product information. So now you know that this group of SUV enthusiasts has other interests that take them to various areas of your site. This has implications both in making decisions on developing new content, new ad products and sponsorship opportunities.
Sounds like site analytics, you say? Well, yeah. In deploying behavioral targeting, you may even be using the same pixel as your site analytics product. What's the difference? In this case, it's your immediate accessibility to information not often in the control of ad operations. It's the speed with which the information can be obtained (24 hours) and it's the flexibility to create measurable segments on the fly. It can also help with managing inventory.
Impact on Inventory
With all the demands on inventory management ranging from demographic targeting to rich media to high sell-through of targeted areas, those of us in ad operations may feel that the last thing we need to add more complexity to this matrix. It just makes us worry more. (Why can't we just turn off the lights at 6 PM and head home?!)
To reduce the complexity, I'm an advocate of creating the equivalent of "no-fly zones" in inventory. By that I mean exclude your most valuable content from behavioral targeting. You obviously don't want to muck up inventory in those areas and cause under-delivery. Focus instead on RON or remnant inventory for the targeting of behavioral segments. Remember the SUV segment? Calculate how many page views they generate when they're wandering through RON (which could include message boards, miscellaneous content, etc). Then, estimate the resulting ad impressions and use that as your behavioral inventory. Call it "premium remnant" if you want. But if the result is to significantly raise the effective CPM of your remnant inventory-- this becomes a winner of a product.
What's the Incremental Value?
Don't look to behavioral companies to help define the value of this targeting product for your site. The truth is, only you know the intricacies of your site, your inventory, your content and the level of demand for the product.
In terms of finding the potential value, take a look at your RON (run of network) and remnant inventory. Here's a sample equation for discovering the potential value for site "X":
- Your site has RON/remnant inventory consisting of 50 million ad impressions
- The CPM of that inventory is $1.00, with a resulting $50,000 in media revenue
- Targeted content on the site sells for $10 per CPM
- Behavioral targeting can be sold for 75 percent of that, or $7.50 CPM
- Convert 20 percent of your RON inventory (10 million impressions) to behavioral
- The resulting media value from the 10 million impressions is $75,000
- You still have 40 million RON/remnant impressions left at $1.00 CPM, or $40,000
- Now, the total value of your RON/remnant inventory is $115,000 with an effective CPM of $2.30, more than double what you started with.
Granted, your site may or may not fit this profile and behavioral may or may not be a slam dunk for you. Like the disclaimers say, "Past performance is no guarantee of future results," "See your doctor if there are persistent headaches or cramps" and "Keep away from small children."
But the indications are that this technique may be a useful part of your arsenal in ad operations.
It may not be guaranteed to double your Mojo, but it's no Doctor Evil, either.
Doug Wintz began his interactive career with Prodigy in 1988. During that time, he pioneered the sales and development of online applications for automotive clients Toyota, Ford and Autobytel, brokerage firm DLJ Direct and grocers Dominick's and D'Agostino. He led the development of one of the first online ad networks for Softbank, managed sales/operations for gamesite Uproar and recently served as VP of digital media solutions for Lycos. Doug is currently founder and principal of DMW MediaWorks, a consultancy in interactive media and operations, with long-term clients that include the market leaders in online health, broadcast television, behavioral targeting and custom publishing.
One of the obvious differences is the type of Twitter Card used. The two most engaging posts were promoted with the objective that Twitter refers to as "app installs or engagements." This means that when a user clicks on the Twitter Card, they are redirected to the app store, where they can download the app onto their iPhone, iPad or other Apple device.
On the other hand, the three Starbucks Rewards tweets were promoted with the objective of "website clicks or conversions," where users were redirected to the Starbucks Rewards web page.
Was the kind of card used the sole reason for the disparity in engagement? Unlikely. In fact, prevailing wisdom says that tweets with images garner higher engagement than those without. So, what gives?
Micro-targeting for business results
Most likely, Starbucks put Twitter's audience targeting to its best use. Did you know that on Twitter ads, you can target not only by gender, region, and language, but by device as well? Simply put, with their Twitter Card redirecting users to the app store, it made sense to target only iPhones, instead of wasting money or reach on other smartphone devices, or even on other iOS devices like iPads and Macs.
Another interesting tactic is the timing of these tweets -- during the post-lunch lull, when everyone is usually jonesing for coffee. What Starbucks banked on was a potential customer scrolling through their Twitter feed on their iPhone, not really able to focus on work after lunch, coming across the tweet, and downloading the app to find the nearest Starbucks. The entire strategy -- from start to finish -- is calculated to align with business objectives.
What can you learn from this?
It wasn't just one factor that made these tweets work so well for the brand. It wasn't just that they were paid for, because other promoted tweets received much lower engagement. It wasn't just the timing of the tweets, as there was no uniformity in engagement in the same time frame. It wasn't even just the type of card used. It was the combination of all these factors, coupled with the customer insights that Starbucks has undoubtedly gained that resulted in success.
The ultimate ROI for Starbucks (and any business) is customers who are ready to buy. Though we don't have access to the number of customers in the time period of this Twitter campaign, according to Starbucks CEO Howard Schultz during their fourth quarter earnings call: "The Starbucks app processed $1.17 billion in 2013, and the company has already processed nearly $1.4 billion in 2014 by the app alone; it's expected to reach $2 billion by the end of 2014."
As social media heads towards a purely paid model for brands, it will be increasingly important to tie social marketing efforts to larger business objectives.
On Twitter? Follow iMedia at @iMediaTweet.