Post Search Ads
Moderated by Misty Locke of Range Online Media She welcomes everyone. The panel will be talking about ways to connect with the consumer after the search. Turns it over to Kevin Lee for a introduction to the concept of post-search ads. He does a quick survey of the audience re: retargeting efforts. Not many in the audience are aware of the “basics” of retargeting. Post search ads are designed to tap into the intent displayed by the searcher’s behavior. The whole idea is to take the intent that occurred and use it to target ads at other points after the fact. You can retarget at engine level, to your own search visitors, and by buying media through networks which specialize in that. Additionally, there is the idea of buying search clicks essentially one click after they leave the original search engine result page.
When a searcher engages in a search, if the engine chooses to have a behavioral retargeting system in place, they can follow and retarget them based on past behavior. Google says they are not planning on using search as a retargeting arena, where the other engines are. The engines have the leverage. Since the search engines see so much traffic, they have a huge pool of ways to retarget. If you are only using your own site to retarget, you will get less traffic. You can do this using ad networks such as Tacoda, advertising.com (both AOL companies), Revenue Science, FetchBack. Getting high scale is always a challenge.
Following the searcher, you can target in two ways: behavioral = graphical target, contextual = link-based. You have to consider when retargeting, why they did not convert the first time they were there. Do you need to change copy, offer/price, and/or landing page? Again, getting scale is difficult here. This is most appropriate for big traffic search sites.
“Post click search traffic” : a “one click away” method. If your top keywords have publisher sites as top organic listings, you can buy placement on those sites. The top ranking sites may have in excess of 50% of their traffic coming from search, so how to capitalize on this? You can target the same searchers, but perhaps those that are choosing more organic results, and then target them with that ad. You can use direct targeting, Google Site Targeting, Yahoo! YPN, other ad networks such as adBrite and Context Web (I think). He also suggests buying links on these sites, recommending a “media budget” for this buying function. This leverages search very well as a targeting area.
“Piggy back retargeting” such as almond.net. They will go into that more when Michael speaks. When you think about retargeting: think beyond the keyword. This is hard since we often think that way in a search campaign. Think about the user – keywords do not fully identify searchers as being in your target audience. The retargeting buyers have ways to leverage the traditional media buying style and methods, and are making it work nicely on the Internet. Kevin asks when will profiles that include search and other behaviors cross from one form of media to another? He states sooner than you may think. Such as moving across media to video games, mobile, IM, and IPTV.
There are not many other industries like search where you can target the audience so specifically. Misty reminds that there is a great value in behavioral re-messaging and how much it can support current activities…this can actually even make media more successful than search alone, which is pretty amazing.
Next is Michael Benedek, from AlmondNet. Thanks Incisive Media, and goes over a quick history of AlmondNet. They have “70+ Million aggregated US/UK users segmented into 40 ready-to-target categories.” Will start by defining behavioral targeting and then move into a case study.
How to define behavioral targeting? “The delivery of ads to a person wherever they go based on their observed online behavior.” Prior actions are the key. Behavioral is one of the fastest growing areas on interactive advertising. As there is more user-generated content and web-based email, these become more difficult to monetize through contextual means, so behavioral gains in importance.
Ways to target behaviorally: “advertiser retargeting” – works best for advertisers with significant scale. Need to find the users once they have left the site. “Inventory extension” – publisher with its own sales force enables its advertisers to reach its audience on other sites. “Data sharing” – publisher provides access to a third party company wand gives site audience data in return for loyalties. Makes sense to allow a third party to cookie your users, and then receive royalties for wherever they go after. Lastly: “post search.” Ads delivered based on purchase-intent data derived from search behavior.
Why is “Post-Search” (they have a trademark on the hyphenated “Post-Search”) so exciting? It all comes down to consumer behavior. 5% of time is spent online declaring purchase-intent on Search engines and other similar sites. 95% of time is spent browsing, so behavioral allows for you to follow them around. How does the technology work? Step 1 is data aggregation. Cookie is placed on user’s browser, and search is recorded and categorized. Step 2 is to identify the users. Step 3, the ad is targeted and delivered. They should not be seeing an untargeted ad. They deliver that ad either directly or in cooperation with a large distributor.
Case study: 600 million behaviorally targeted ads delivered over a three year period revealed that CTR for BT are similar to CTR for untargeted. Clicks originating from Post-Search behavioral targeting converts 5-10 times better. Benefits: users win, “we all win.” Privacy is maintained, since the user’s info is not shared in order to deliver BT. This allows SEMs and Agencies to target people during the “other 95% of the online time.” Data providers can earn an incremental revenue stream. Media owners and ad networks get to maximize advertiser ROI and CPMs. “What transforms difficult to sell inventory into valuable inventory is data.”
Is there a secret sauce? Not really, ads are targeted based on a recent demonstration of intent. Problems exist with contextual: person reading about insurance fraud is not likely interested in buying insurance. Just visiting a NYTimes financial section page does not mean that the person wants to open an online brokerage account. Behavioral targeting tends to do better at avoiding errors like this.
Next is Dave Carberry from advertising.com’s Out Search, and a writer of popular childrens’ books. Ad.com reaches about 88% of the internet population. Talks about the advantages for the advertiser. He shows how their post search system operates through the use of pixels. It is very important to re-engage your consumers after they click. He loves cookies, although 15-20% of the marketplace may actually delete theirs’. The desired conversion should be the key, and if that doesn’t happen, they can retarget and try again. The “behavior” is the searched keyword. The typical window is 14 days, in order to retarget. You can also target 5 days, or even 30, as long as the user has allowed the pixel.
The information/data is anonymously tracked and non-personally identifiable. They are not collection information, rather the behavior. They can also overlap geographical areas or demographics based on the desired audience size. They can provide custom messaging from multiple touch points. The segmentation can be both user and keyword, and this broad segmentation keeps the retargeting pool wide. Shows some sample segmentation categories such as automotive, sports, weddings, etc.
The key players are AOL, Yahoo, MSN, and Google. Last quick tips and tricks: 14 days retargeting window is the ideal time frame. Additional product branding in the messaging = a higher potential for conversion. Use frequency capping to not “overdo it.”
Next up is Richard Frankel who is a senior product director at Yahoo!. Will do a deeper dive into how some of these retargeting products work. Yahoo! has an “audience advantage.” They have a very wide reach, and not only for search. Since they can look across their network for behaviors, they can determine even more information. 70% of Internet users visit Yahoo! They have two main behavioral products “Engagers” and “Shoppers.” These are geared towards different times in the buying cycle.
“Yahoo! Engagers” is more for brand awareness and allows for a deepening of the engagement in order to build brand preference. “Shoppers” is geared later in the buying funnel. How it works? They analyze predictive patterns for ad response in 350+ product categories. They have to understand the varying product purchase cycles, and analyze these patterns to find out what consumers are really doing before making their purchase decision. They build models for each category in both “Engagers” and “shoppers”, score each of them daily, and target ads to users who get the highest “relevance” scores in the targeting categories searched.
He feels that this process of “response prediction” will be the future model for this area, as it works very well for Yahoo! They have seen fabulous results, but he knows they can do a whole lot better. For example, someone wants to target people that want digital cameras, but then show “generic ads” that don’t narrow the targeting with the specific messaging. This leads to their “Smart Ads” product. It dynamically personalizes/generates the message specifically targeted to the interest. Shows an example of a highly targeted message, which they have found generates 2-3X CTR and 9X the conversion rate.
Steps in the process are similar to others’: 1: “Smart Target” (identify the user). 2: “Smart Message” (personalize the message around the user). 3: “Smart Action” (use a specific actionable page – they do not want to lose the user because of brining them to a home page or some other non-conversion-focused page).
(This is live coverage of SES San Jose 2007, and some typos or grammatical errors may exist. If you were a panelist and you would like something clarified, please post in the comments or contact me through the system)