I’m taking a new approach with events and attempting to do more in-depth coverage of a couple of what I feel are the best, most interesting, most innovative or least discussed sessions at the conference. Since contextual advertising is an area where I have little to no experience, I found the panel fascinating – and certainly Microsoft’s announcement of a contextual ad product on the Microsoft network of sites made this panel even more intriguing. What follows are excerpts of interest from my notes, organized as best I could manage:
Panelists: Anton Konikoff from Acronym Media (BTW –Β this isΒ very odd), Lora Parker from Range Media, Gopi Kallayi from Google AdSense and Natala Menezes from Microsoft AdCenter
Anton Konikoff was up first. He works with a lot of large companies, including Accenture, WebMD, Sirius, ADP, Nokia, Siemens, Four Seasons and many more. According to Anton, there are four prevalent types of online activity:
-
Searching using keywords
-
Clicking on hyperlinks that contain keywords
-
Looking for keyword matches in written content
-
Creating content using keywords
He claims that “there is a definite keyword theme in everything we do online.” I’m personally thinking that a lot of blog reading, content consumption and email isn’t truly keyword driven, but he’s certainly right that, at the least, a lot of online activity centers around keywords. Anton also walked us through some examples of basic contextual ads:
_
The “Ads by Google” section on the left shows contextual ads for Pink Floyd
_
_
_
These ads appear at the bottom of many InformationWeek articles and are served by IndustryBrains
_
Anton notes that contextual search does have some weaknesses – it generally does not perform as well as search advertising (which shouldn’t surprise most of us in the search business). However, it does present some exciting elements:
-
Provides an additional layer of customer intelligence that helps us to better understand customers use of a keyword
-
Helps with βsocialβ context and βverbalβ context
-
Context can redefine messaging β different visitors to different sites respond differently
So when is contextual appropriate? According to Anton, it’s when it:
-
Gets around the search inventory issue
-
Provides access to everyone who reads online (not just those who search)
-
Assists with awareness building (I have my doubts here, personally – I don’t know how many people really SEE contextual ads as they read other content)
-
Fits with direct responseΒ and providesΒ flexibilty with CPM/CPC/CPA pricing models
One of Anton’s most brilliant tips, from my perspective, was his suggestion that savvy advertisers can “supercharge” press releases by targeting a term/phrase that is unique to a single press release or product so that call-to-action ads can appear alongside any mention of the news. This is the kind of clever thinking that makes these conferences worthwhile, IMO. That tip alone is probably worth the entire price of a trip to San Jose for the right folks π
Anton continues and mentions that for Acronym Media, which does a lot of work in the contextual space, but also provides other SEM services, 40% of the traffic they drive to their clients’ sites comes from contextual. In certain countries with low search inventory, this number goes up to 90%.
Anton notes that all of the following provide contextual ad services:
-
Google AdSense (the leader in the field)
-
Yahoo! Publisher Network
-
Industry Brains
-
Quigo
-
Pulse 360 (formerly Kanoodle)
-
Ask (though this is largely limited to IAC properties)
-
Context Web
-
Vibrant Media
He also provides a bucket-load of tips for optimizing campaigns:
-
Use fewer keywords in each ad groups β use many ad groups with very tight targeting and more granularity (which allows for better tracking)
-
Use Exact and Phrase Match
-
Always separate search from contextual for tracking (this may not be easy, as engines don’t always do a good job showing this)
-
Track on a site level where possible β track based on referral source (so you can exclude non-performing vendors)
-
Leverage your best converting PPC search terms (this is not a literal translation but it’s a good starting program)
-
Look at where banners perform best and target those sites first
-
Negative keywords β add them over time as you see lower value and obviously irrelevant terms
-
Site exclusion β as above
-
Your rep can help you to get rid of sites for you, so build up relationships at the engines if you’re doing contextual advertising on a large scale
-
Ad Copy in contextual must be more persuasive than the rest of the page copy (often more so than search ads)
-
Contextual clickers are usually NOT as far along in the buying process, so the targeting point is earlier (this means you can capture and brand earlier, but should expect a lower conversion rate on average)
-
Create different calls to action based on the breadth and depth of the source of the ad β niche site calls for different ads than broad news site
-
Most publishers opt for only a few ads on their sites, so it’s critical to be in the top 2-3.
-
In testing from Anton β the actual rank is less important β i.e. #2 and #3 get almost as much traffic in contextual as #1
-
Relevancy in AdWords is determined at the AdGroup level, not the CTR
-
Work to avoid negative association, but be aware that there will be bad stuff out there. For example, Anton’s client β Four Seasons βΒ is worried that contextual will damage their brand association. They only want to be near premium/luxury travel content, but because of this, they miss out on a lot of traffic and branding opportunities.
A major concern that many folks seem to have centers around click fraud. Anton has some recommendations here as well:
-
Sign up as a publisher and experiment – the knowledge you gain by running contextual ads will make you a much smarter buyer of ads
-
Learn all the options and tools available to publishers β launch AdSense on your own sites
-
Look at sizes, formats, colors β what works to get you the best clicks?
-
CTR is largely irrelevant – .033% CTR is terrific β you don’t want to increase CTR necessarily. Look at above-the-fold impressions, conversions and ROI
Anton mentions a couple of sample clients that he’s worked with, which is exceptionally engaging. The first, Free Ringtones Galore (which I can’t even find because ringtone SERPs are so spammed up with junk) was able to reach an audience that was far too expensive to approach in search ads through contextual. The second, Humana Medicare, targeted seniors in particular and climbed from an initial conversion rate of 0.5% to over 3% by eliminating negative keywords and careful site targeting over the course of several months. Overall, contextual for Humana delivered 10X the impressions of search advertising at a 30% lower cost. To me, this shows just how much work is required to stay on top of contextual ads, but clearly the return is worthwhile.
Anton kindly invited anyone to email him with questions – anton at acronym.com
The next speaker, Lora Parker from Range Online, presented a variety of case studies from the world of contextual.
Consumer Electronics Client β global brand with no brick-and-mortar.
-
Focused primarily on brand terms
-
Allocated 5% of total budget to contextual
-
25% higher ROI on contextual than search
-
35% of total campaign impressions
Education client β not a well known brand
-
Launch with limited test β only top performing terms
-
1% of total budget
-
13.2% higher cost-per-lead than search
-
10% of total impressions
-
Conversion rate β 50% lower than search
-
Content is seasonal and will run only during peak seasons (when meeting cost per lead goal)
Financial Services β well known brand (leader)
-
Goal β drive transactions at $30 per transaction or less
-
Allocated 10% of total budget
-
Cost was 50/50 between branded vs. non-branded
-
18% of transactions were completley new customers (search at 23%)
Retail β Luxury shoe retailer β goal to drive online sales
-
Launched with brand and high non-brand performing terms
-
5% of budget allocated to content
-
Increased impressions by 4X
-
Store locator jumped 15% from contextual (offline metric)
-
Campaign ROAS β not met, so program was discontinued (they think they can refine as Google, et al get better)
Technology/Software but not well known brand
-
Launched only brand terms and some high performing non-brand terms
-
10% of total search budget was allocated to content
-
Ended with slightly below ROAS goal
Travel β not a well-known, but an up-and-coming brand
-
10% of search budget to content
-
Tested in all categories β hotels, flights, etc.
-
50% of total impressions from content
-
Conversion rate lower, but CPC also much lower
-
Top performing category is only one left running (in air)
Average Numbers:
A quick note – Anton says during Q+A that the average numbersΒ are useless (though I find them very compelling as a way to both introduce potential players to the contextual search world and to compare against any campaigns we may run). In his opinion, however, the only real expectation is lower CTR – everything else is up for grabs depending on the campaign.
Lora kindly provides her email as well and offers to take follow-ups β Lora at RangeOnlineMedia.com.
Gopi Kallayi, a (maybe “the”) product manager for Google AdSense, was up third. Sadly, his presentation included no Powerpoint and lacked any quality content about the subject. He noted that Google had shared a total of $3.3 billion in advertising revenue last year with publishers then talked about two stories of his own personal experiences of seeing relevant contextual ads for pot washing faucets and diapers for birds. While entertaining, I certainly didn’t learn anything and felt frustrated that Gopi had no deck prepared or relevant data to share – particularly since Google is the market leader in this space and so many folks have questions about the ins and outs of AdSense optimization. On the plus side, Gopi did allow more time for Q+A, which proved quite valuable.
Gopi was also the only presenter to decline to provide an email address, even when specifically asked for a support email during Q+A.
The final presenter was someone about whom I cannot claim any degree of objectivity. Natala Menezes of MSN AdCenter is not only a colleague, but also a great friend and one of my very favorite people. It appeared to me that many of the folks in the audience knew or had known Natala previously and also had positive opinions of her prior to her presentation – she was certainly warmly received (which is good, since I’d probably have to shoot nasty looks at anyone who was mean to her).
Natala announced some interesting news from Microsoft – primarily that they will be offering contextual advertising for advertisers on the Microsoft AdCenter platform on many MSN properties that were previously closed to only large advertisers. The ROI and quality here is expected to be very high, while the costs will most likely be low, at least at first. I suspect there will be a number of folks who are able to very effectively monetize this new inventory availability, at least in the first few months.
One very controversial part of the announcement, however, is that Microsoft, like Google, will be turning on content network advertising by default – a move that I can understand, but may not provide the best experience for advertisers. Basically, let’s say you are advertising on AdCenter today for a search term – after next Wednesday, you’ll be automatically included in the ads that appear on the content network. To Microsoft’s credit, advertisers do have the ability to opt-out until this Sunday night, in advance of the launch and can opt-out at any time after that (of course). They’re also sending out emails to all advertisers indicating the change.
Despite these concessions, many folks in the room still seemed surprised, if not fully taken aback. There was considerably complaining by multiple parties and a general grumbliness about the move, though, to be fair, much of this criticism was equally directed at Google for auto-enrolling all advertisers in Google AdSense and forcing an opt-out.
During Q+A, Natala was asked what properties would be included in the contextual rollout, and mentioned that in addition to MSN sites (like autos, finance, travel, real estate, health, etc.), Facebook, Digg & the Windows Marketplace website would all be included. She also was asked about a rollout for Canada, but noted that Microsoft has no timeframe for that launch. One excellent question asked – why is Microsoft selling these ads to smaller advertisers if they currently provide them to large, display advertisers. Natala responded noting that in many cases, even the high CPM rates were not as optimal a use of the space as targeted, contextual ads – fascinating stuff.
Natala also provided her email address for feedback β natalam at microsoft.com or adccfeed at microsoft.com.
To me, this session was particularly illuminating. I’ve never before considered using contextual ads, as my experiences in the past with them have been horrifyingly bad (low CTR, no conversions, high costs and usually the result of me forgetting to find and turn off contextual inside the AdWords console). After this, I’m definitely feeling a lot more educated and ready to take on the process.
Please let me know in the comments (and with the thumbs) not only what you thought of this particular post, but if you enjoy this coverage format for events.