Thursday, December 10, 2009
At SES Chicago, a member of Google's product management team routinely asked me if there were particular aspects of the AdWords interface that we ("power users") would like to see improved.
Sometimes I'm at a loss when it comes to questions like that. There is so much going on in an interface like that, specific suggestions can slip my mind... and in any case, in the beta process for the new interface Google must have had to weigh thousands of pieces of input anyway (so it's really the weighing that is the issue as opposed to any lack of cool feature ideas).
To be a smartass, I certainly might have pointed to the campaign setup phase that hides the need for us to enter separate (or no) bids for the content network or to more clearly offer advertisers the ad "rotation" (as opposed to "optimize") option. Google makes more money when advertisers forget to address these settings, so I basically don't expect them to change the state of affairs that amounts to "usability in reverse" on the campaign setup. It's not unlike hard-to-read credit card statements. The less people know, the more the company makes.
Asking for a small feature change misses the point, if there are significant areas that affect campaign economics negatively, that haven't budged.
So then it hit me -- we still haven't seen any movement on the issue of differential bidding for the "search network." George Michie recently reiterated what is basically a unanimous feeling among advertisers: that we need "syndication controls."
To refresh your memory, there are essentially three big channels available to AdWords advertisers, with many controls underlying two of them. (1) Google Search; (2) The Search Network; (3) Content Targeting.
Google Search is self-explanatory. Content targeting is a network of online publishers, from the biggest to the smallest, who participate in the "AdSense" advertising program. For advertisers, there are two flavors of this program: automatic placements and managed placements. Both offer stupendous levels of control for the advertiser, including the ability to set specific bids.
The search network is other search engines who display Google AdWords ads much the same as they appear on Google Search: AOL, certain Internet Service Providers, Ask, metasearch engines, etc. But in addition to that there are "navigation like" properties. Google publishes no formal exhaustive list of these partners.
This is where it gets weird. You can opt out of the search network. But who would want to, as it is a decent chunk of good traffic that typically far outstrips the buying intent of things like display ads, etc. Unfortunately, on most accounts, it universally performs worse, with CPA's (costs per acquisition) coming in 20-30% higher. And you cannot bid on the inventory separately or opt out of specific partners or segments, something you can do extensively in the other two megachannels within AdWords. So you're either all in, or all out. That's strange, and it hurts advertisers. I join Michie in his "rational plea."
Labels: content targeting, google adwords
Saturday, May 02, 2009
What's better: 100 ad sales people in your ear all trying to get you to buy subpar inventory? Tendentious studies trying to prove "brand lift" from ads even when they don't seem to convert?
Or a network that is quietly optimized and improved over the years, one that adds control and opt-out features including separate bidding, domain exclusion, and reporting on content types, such that it performs -- yes -- just as well as search on a CPA basis?
Why, the latter of course. Brand lift is real, and lattes with ad reps can be fun, but when I talk to my clients, for the most part, they're looking at AdWords like direct marketers and want measurable results.
Google's White Paper on CPA Performance Trends on the Google Content Network, based on data from 25,000 accounts, shows that:
Additional notes, from the fine print:
- On a CPA basis the Content Network performs about equally to search;
- Content provides about 20% of the conversions in the median account;
- Two account controls improve performance in particular: (1) site exclusion; (2) Conversion Optimizer;
- Canada is getting the best deals on content on a CPA basis; the worst return is reported for the UK market. In line with past conclusions about these markets, I assume this is because Canadian advertisers are stingy with budgets on average, and not enough have entered the auction. In the UK, account management practices are sometimes shoddy, and anectodally, UK Google reps have encouraged advertisers and agencies to overbid, so until these practices improve, CPA's will not. The US places second behind Canada in content CPA performance, in keeping with expectations again: in the US there is widespread account optimization; many accounts today are actively managed and scrutinized as opposed to "set and forget".
I have questions about the methodology, as follows:
- Image and alternate format ads were included in the study, but 86.9% of the spend under study came from text ads.
- 90.3% of the spend under study used classic content targeting using keywords to trigger placements. Presumably the remainder used the new custom Placements options.
Despite some open questions about the methodology, based on accounts I see these days, I roughly concur with the conclusions. Though I think it is essential for advertisers to "do some special things" to succeed with content, including bidding lower in many cases, the CPA and ROAS trends have been quite favorable.
- Beyond these account controls, how many advertisers are doing anything special to actively improve performance?
- Is the sample, based on a billion clicks and 70 million conversions, random? It doesn't sound quite random, but it does sound like anomalous accounts were tossed out, and a high number of accounts (>500) were aggregated in any given vertical.
- Aren't the conversions a jumble of registrations, sales, leads, and actions of many kinds, jumbled in together?
- Will a smaller study become available, one that looks at conversion revenue divided by conversion cost (ROAS or Return on Ad Spend) for those accounts that pass revenue through to the reporting?
- What proportion of these accounts have been optimized by a Googler putting in research hours to select specific placements, tweak keywords, set up separate ads, etc.?
Based on the huge volume of accounts studied and the compelling way they've presented this data, I have to say this is an impressive study from Google. And they slipped their call to action in (no Buy Now!) in pretty subtly: advertisers in Canada, the United States, and Italy should look again at enabling content, or bidding higher on it. Those with broken-out content-only campaigns should consider using Conversion Optimizer to improve CPA.
Labels: content targeting
Wednesday, March 11, 2009
My, how well organized Google was to announce their new behavioral ads in content targeting offering.
To the public at large, VP Product Management Susan Wojcicki provides the official explanation of why digital advertising is good, how Google tries to make the ads relevant, and what behavioral targeting is going to entail. Above all, this part of the announcement scrupulously covers the privacy principles and the user control and opt-out elements of Google's advertising technology.
For AdSense publishers, a good-news announcement that advertising will continue to reach more relevant audiences, and a brief allusion to settings in AdSense accounts that deal with the new capability.
And finally, the piece de resistance: what about the AdWords gang? For the paying customers, the advertisers, a post on the Inside AdWords blog explaining that this program will be in beta for several months and that it will be expanded "later in 2009." Product manager Aitan Weinberg writes about this offering "helping you to better reach your campaign goals at scale." Concisely put, and that is exactly what we're looking for out here in advertiser-land.
It's hard to imagine any company covering the communications bases as well as this for a product that won't be fully rolled out for 4-6 months. Google, if you were expecting a major firestorm of controversy about your announcement, I have to apologize in advance. It just isn't going to happen.
Labels: content targeting, google adwords
Monday, March 31, 2008
A little self promotion never hurt anyone, right?
That’s the attitude Google adopted in recent months as they began aggressively promoting their content network, albeit with some help from their friends.
Google feels strongly about their content network – namely, content targeting’s ability to gets Google AdWords advertisers’ names out quickly, easily, and potentially at a lower CPC than a traditional AdWords campaign – and they want everyone to know about it. That’s why Google advertisers have recently been direct-marketed with “success stories” about happy clients using the content network.
It’s particularly interesting to note that today’s Google success stories are pushed out to advertisers right inside the AdWords interface, in an announcements area, so it seems almost part of the navigational experience.
Take the SEM agency SearchRev, for example. They’re gushing over Google AdSense after seeing retail clients increase revenue by 25 per cent since implementing it. SearchRev used the content network to test out a few “general” type keywords that didn’t include clients’ brand names. As a result, retailers successfully introduced themselves to new customers who hadn’t otherwise encountered them before. Even though most of those newbies didn’t click through the ads, it didn’t matter; many of them ended up searching for the retailer later on and joining the company’s growing list of keyword search targets.
Since SearchRev specializes in SEM, and thus qualifies as a “third-party agent,” eyebrows may raise at the thought of them partnering with Google. Does a third-party endorsement look suspect when it’s done within the AdWords interface, providing the agency with exclusive exposure?
In theory, this new reciprocal promotional arrangement with SearchRev could make Google a hypocrite. But can we really fault them for finding a creative way to advertise? They are a corporation, after all. More importantly, recent improvements to AdSense suggest they’re listening to customer demand and backing up their self-promotion with a quality product.
Despite its relative success, Google’s content network developed a somewhat shady reputation among advertisers in recent years, with rumours of “built-for-AdSense” booby-trap sites surfacing. Google responded with several positive changes to the content network over the last couple of years. This includes placement performance reports and various new flavors of site exclusion.
Sure, Google’s running the show a bit differently than in the past, but given their growth imperatives today, it’s unsurprising that they’re working harder to push their product.
Labels: content targeting
Friday, November 09, 2007
Nag incessantly, and ye shall receive.
I don't see this on Inside AdWords, so it must be new. I just stumbled on an opt-out functionality for the content network, being beta tested currently.
Two types of "content exclusion" have been added alongside site and page exclusion options: (1) topics; and (2) page types. Under "topics," advertisers will have the ability to opt out of certain types of content, like "edgy content" and "death & tragedy." More benignly, page types you can exclude include "image sharing" and "domain ads."
Wow. And in the opt-out interface, Google is really hitting me over the head with the recent conversion stats on some of the different types. In the campaign I chose, domain ads did just fine today - outperforming the campaign as a whole, in fact, from an ROI standpoint. There's no point in disabling something if it actually works, obviously!
So be cautious when this launches. If content is working fine for you, of course you'll want to avoid unduly limiting your exposure in the universe. In fact, if you can exclude unwanted content types, it might be wise to raise your bids on the rest.
For advertisers concerned about showing up in unwanted places, this is great news. Also great will be the fact that we'll be more armed with the facts about ad performance on different types of content. Heat gives way to light.
Labels: content targeting, google adwords
Monday, June 04, 2007
Today Yahoo! announced quality-based pricing for clicks coming from distribution partners in the sponsored search and content match programs. Similar in intent and implementation to Google's "Smart Pricing," it will automatically discount a publisher's click revenue (and the amount the advertiser pays for that click) by "a certain percentage" based on traffic quality as measured by conversion rates and other factors.
This systematizes the same process that has taken place with interactive ads for years: advertisers run tests, find out which sites convert well, and reprice their offers when they re-order, lowballing the low quality publishers.
This is a much-needed initiative for Yahoo!. Although it will initially dampen revenue, it will put Yahoo's relationship with advertisers on a solid foundation, allowing for future growth. I've witnessed first hand the slow increase in the ROI graph on Google's content network, to the point where we now feel confident bidding on this inventory in almost every advertiser account (although still lower than we bid for search inventory, usually). Smart Pricing has been an important part of this picture, as blunt an instrument as it may seem to some publishers. It's worth noting that by taking away ill-gotten revenue from bad publishers, these networks can afford to pay out better to quality publishers.
This initiative will further cramp click arbitrageurs' style, though not all forms of arbitrage will be negatively impacted, because not all publishers are part of the same content network (did you know that?) so different rules might apply. That being said, these distribution partners that are classified as falling under "sponsored search" appear to be subject to the new quality-based pricing; however, nothing some select partners won't be whitelisted or made exempt from the policy.
One possible criticism of such policies (I called this an "actuarial" approach to content pricing when Google rolled it out 38 months ago) is that it's a way of blunting criticisms of fraud-prone distribution partners without actually offending the partner network by taking direct action. In other words, it's an abdication of past responsibility for click fraud in the network, while trying to quietly rectify the problem. A further indication of a gradual acceptance of responsibility for the quality of the partner network is in point 17 of the FAQ's: "Although we do not currently offer the ability for advertisers to opt-out of particular sources of traffic, this is a feature that we plan to begin offering later in 2007." Good thing, because for some partner traffic, as far as many advertisers are concerned, the only smart price is zero.
Labels: content targeting, contextual ads, yahoo search marketing, ysm
Tuesday, April 17, 2007
A few weeks before the recent announcement that Google is buying DoubleClick, DoubleClick announced it'll be developing a "NASDAQ-like exchange" for the buying and selling of online advertising. Proof positive that the online ad space continues to heat up, rather than diving into an irrational "bust" as in 2000-2003.
Wait a minute, though, you may be saying. Aren't ad networks and the way ads are bought and sold today somewhat "market-like"?
Not as much as you might think. Check out Google's current contextual offerings, for example. As a publisher, do you think you can signal to the advertiser using an "ask" price? Do you have much control at all if you join one of the various ad networks and let their system allocate inventory, as opposed to using your own inside sales force? Nope. And what about the ability to stand out as a quality publisher amidst the crowd of remnant type inventory? Again, tough to do. So, tough to monetize to your fullest potential. Because the current market maker is not facilitating a true market. In that sense, DoubleClick's new promised offering, and Google's acquisition of DoubleClick, could constitute a significant step forward in online advertising efficiency.
DoubleClick won't be the only one working on a great way of putting online ad buyers and sellers together. ContextWeb, quietly licensing its matching technology since 2000, and more recently, acting as a media middleman in its own right, is working on a new "name your price" functionality in a new, highly automated ad marketplace system. I had the chance to walk through a demo yesterday with CEO Anand Subramanian and VP of Business Development Jay Sears. While the release isn't slated for a couple of months, publishers are already being targeted for a beta signup. The hook, "Ready to Make More Money than AdSense?," is not new to publishers, admits Subramanian. "Typically a publisher will become dissatisfied with their AdSense earnings, will install code from a competing network, but that will be even worse, so it's right back to AdSense," he says. But that's because these competing networks don't present a credible alternative. They represent the old school of what ContextWeb calls "Yet Another Ad Network," with fewer features, not a true marketplace, and no critical mass of buyers and sellers.
The company won't yet disclose the full range of features of its new system. To me, it seems to address the combined needs of advertisers and publishers very well, leaving neither the "prime inventory" nor the "long tail" unaddressed. ContextWeb's quiet long-term presence in the space seems to have given the company a wealth of ideas. A decent revenue stream from its current agency relationships, plus venture funding from, among others, Draper Fisher Jurvetson, has given them the long view needed to build a better platform.
If you're a publisher of quality content, here's hoping the days of "Yet Another Ad Network" will soon be a thing of a past in your balance sheet.
Labels: anand subramanian, content targeting, contextual ads, contextweb, doubleclick, google, jay sears
Friday, April 13, 2007
At Search Engine Strategies New York this week I had the opportunity to model a panel on the latest with buying contextual ads with particular focus on the top contextual programs through Google, Yahoo, IndustryBrains, and a couple of others.
One of the most fascinating was by a major cable television network that uses low-cost, broadly-based terms to drive traffic keying on all kinds of current pop culture related content. It's an actively managed campaign that requires constant innovation to stay ahead of the curve. What's so interesting about it is how cost-effective this is as a way of generating buzz, if you get the tone right. It's basically using online media to deflect user attention back into the media vortex, in a subtle manner.
The two main economic benefits stated by the panelist were:
* Low-cost awareness-building for their flagship hit programs;
* Traffic arbitrage, sometimes breaking even or better on the ad inventory they show on their own sites.
Let me repeat that again. In a world where we've suddenly been conditioned to believe that PPC arbitrage is wrong, this marketing executive unabashedly admitting to doing it.
This paralleled my own presentation the same week, on how Google currently measures site and landing page quality. In short, I wanted to make clear that you could technically call a lot of the media companies advertising on Google "arbitragers," because they know the rough CPC's and effective CPM's on their ad campaign with Google, and they know the rough payback on an impression basis from their already sold inventory. So in fact the distinction is not a literal one, where you point a finger at someone making a profit on a media buy/sell and call it evil; rather, Google's quality scoring formula aims to disincentivize certain advertisers from offering deceptive or particularly annoying user experiences as defined by user input and user behavior. In an upcoming column I'll look more at the distinctions between "arbitrage," "nearbitrage," and "garbitrage."
In short, to have an exec admitting to arbitrage is not scandalous. If they have advertising on their site, and are buying ads to drive to that site, anyone could have figured it out anyway. That's what they're doing. As a bonus, they get cheap promotion for their TV lineup. And some publishers get paid too. Win-win-win.
Labels: click arbitrage, content targeting, contextual ads, search engine strategies, ses new york
Wednesday, April 11, 2007
You know you're doing something right when a speaker hanging out in the green room here at SES NYC unplugs his laptop and wanders over to show a fellow marketer a marketing campaign, as sort of a 15-second case study. This Coffee Fool appears to be a fairly heavy user of contextual advertising at the moment, showing up adjacent to GMail and such. (Not a really surprising semantic match given that you probably type a lot of messages like "meet for coffee?" in email.)
I'm not sure what "coffee secrets" are really "exposed," but hey.
Disclaimer: I don't endorse or know much about their marketing methods. I don't know the company. I just noticed them.
Very important disclosure: I was not paid to post this!
Labels: coffee, content targeting, contextual ads, google adwords
Monday, February 26, 2007
Seth Godin wonders whether media buyers are right to pull ads off Google's and Yahoo's contextual networks because of how loosey-goosey they are with their approach to placement -- they match ads to pages, rather than allowing the advertiser "channel control."
While it's true that Y!&Goog would benefit from better sites joining their networks, I agree with Seth that being so afraid to show your ads on "Joe Schmo's Sports site" could be doing the client a disservice.
I thought we already had this debate. In the early going, lot of us were critical of the contextual ad programs for a number of reasons - mine was simply the poor performance, fraudulent or crapulent publisher partners, etc. Others in the biz, with more of an agency bent (and most likely to cheer for Quigo, Sprinks, etc.), demanded that Google's content targeting allow more direct control of what websites ads show up on, as opposed to forcing advertisers to accept the open-ended "smart matching" concept that used semantic technology to match ads with content.
So... first Google responded with site exclusion. Then, they released a site-targeted flavor of content targeting, in a parallel program. That as a direct response to these agency-style demands. Site targeting allows you to browse a menu of sites, add them to your list, and only show your ads on them.
I monitored ads running in both flavors for several months. A funny thing happened: the old "flawed" content targeting program got better, and my approach to managing those campaigns improved. The ROI came in line with search. Meanwhile, nothing on the "site targeting" side was converting. The performance was much worse.
At a couple of conference presentations I guessed that this is in part because computers do a lot better job of matching my ads against a million potential candidate pages than I possibly can in scanning down a list of 50 so-so potential publisher targets. You settle on 20 or so of these sites, then become obsessed with spending the full budget on just those. They convert poorly, so you've overspent on this handful of websites. That's a fairly typical scenario.
In short, because of computers aiding in the matching, classic content targeting offers more efficiency, as the systems get perfected.
Seth, both the intuition and the data point towards there being nothing inherently wrong with Google's approach to matching ads with content. No, the program isn't perfect, but placing high-CPM ads on big brand sites just because I want to appear respectable isn't exactly a challenge. It's more of the same: take too much of the client's money, and waste it, and claim the blue-chippiness of that approach as a benefit.
Both approaches -- the finicky put-me-only-here approach, and the "ROI-or-else" approach -- work in the marketplace. For very different reasons. Funnily enough, Google now offers two parallel programs to suit different ad buying constituencies, and are working on rolling out multiple ad products down the road, to keep everyone happy.
Labels: content targeting, contextual ads, google, google adsense, long tail, quigo, seth godin, yahoo
Search Engine Land has it. Expect more Google announcements re: click transparency in the near future.
Labels: content targeting, contextual ads, google adsense, google adwords
Tuesday, February 13, 2007
At least one panelist (David Szetela) on the Ad Program Strategies: Compare and Contrast panel here at SES London, argued that while he liked the control of Google's Site Targeted flavour of content targeting, he wasn't so fussy about the CPM-based pricing model.
Coincidentally, today Google is announcing that they're beta testing a CPC-based version of Site Targeting, so advertisers who like the idea of paying by the click (that's pretty much anyone who is used to PPC as it's implemented in the rest of your account, and on YSM and MSN adCenter too) can do so.
Seems like advertisers just don't like switching to CPM-based thinking inside of a CPC-oriented platform.
Related Traffick posts:
Note, even if we go entirely over CPC's, it's always a snap to track the associated cost in CPM, if you're into apples to apples comparisons. In previous posts I had argued that even in traditional CPC-based content targeting, some of my favorite campaigns were successful precisely because of the price - which was effectively about $0.25 CPM.
Labels: content targeting, contextual ads, cpm, google adwords, paid search
View Posts by Category