Tuesday, April 13, 2010
As Twitter moves to pilot its first experiments in monetization, it might be interesting to speculate on its prospects for success. To help, I'll go through some of the elements of success and failure that have been proven in the last twelve years or so of online advertising experimentation. Without all of these elements being in place, ad-supported models have tended to fail.
1. Large enough audience to matter. Wrapping some ads around content or functionality geared to a relatively small audience is tricky on a number of levels. First, no one in the press cares, and investors don't care. Most importantly, advertisers and agencies don't care, since there's not enough to buy, so you get lumped into remnant or at least underpriced network inventory unless you've got a really smart little sales force. Second, any hiccup gives you a greater chance of killing the golden goose of whatever you wrap the ads around. Third, you lack statistically significant data for testing and refining, so it's hard to perfect. Fourth, related to the third point, dipping a toe into the water becomes difficult. Large publishers can run tests without alienating anyone as they test the model in a small sliver of the content.
2. Targeting by keyword. Publishers and ad mavens have bent over backward to insist that targeting can be based on concepts, personalization, demographics, and factors other than keywords. Even Google, the King of Keywords, began fairly early in its attempt to paint the keyword as only one sub-facet in the global effort to better align advertising with user tastes and intent. (Bonus: that effort to blend into the woodwork might have helped Google in court if trademark and patent lawsuits really started to escalate out of hand, or if they started losing cases so badly that they'd need to substantially revise their business model ahead of schedule.) Deny it all you like, but keywords still "click" with advertisers. Users like them too, because it's a way of seeing relatively relevant ads without feeling too creeped out. Keywords triggering relevant text ads and offers are the display-advertising-in-content cousin to permission marketing as it was conceived by Seth Godin for email. Somewhere, a line can get crossed. Keywords do a really great job of helping advertisers and users connect without that line being crossed as often.
3. Doesn't get in the way, or even at times enhances the experience. Advertising is a necessary evil to some, but to a substantial part of the population, it's a buying aid or even a cultural experience. Glossy ads in fashion magazines are part of the "art" and "positioning" and are seen as less intrusive than advertising that really "gets in the way" of reading an article online. The same goes for billboards by the highway: an eyesore to some, they're a part of cultural history to others -- and hence, provide free buzz over and above the advertising cost. Burma Shave was before most of us were born, but chances are, you've heard of the roadside signs.
4. Is in a place online that people willingly go to or are addicted to, rather than being an app that is a bit cumbersome to use, take-it-or-leave-it, overly incentivized (paid in points or cash to "surf,"), or weakly appreciated but maybe a flash-in-the-pan. Related to this, the user base has to understand what the owners plan to do around advertising and what kind of "trade-off" they can expect. Do they get involved in using something for one reason, then find it's infected their user experience or device (i.e. "scumware")? Or is the format and the trade-off relatively transparent?
5. Isn't susceptible to "banner blindness". For the time being, we can consider this one relatively unimportant, as initially, enough advertisers will be lining up to try new things where the audience is big enough and attention can be grabbed. But performance marketers are turned off by ads that don't perform, and historically these types of ad formats have had limited upside when compared with personal, anticipated, and relevant communications (especially when the latter are connected with keywords). You can be "big" with the support of brand-building advertisers, but with the approval of direct marketers on top of that, you can be huge... because then any advertiser, large or small, can justify it to themselves or to someone on their board of directors. And agencies too can come up with those justifications.
To look at some quick examples:
- Intrusive or oversized display ad formats -- leaderboards, "popovers," garish animations, etc. -- have had mixed success. They've driven online advertising to a degree, but somehow got surpassed by little old search, despite their reach. That's because they fail on counts 3, 4, and 5, and aren't even all that great on 1 and 2.
- Weird apps like Pointcast, eTour, and Gator eventually fail because people uninstall the apps, don't install the apps, etc. Performance is uneven and users squeal. To incentivize users to do things they wouldn't otherwise do, you either deceive them or pay them too much (thus killing profit). Fail all around.
So how about Twitter? Twitter's scheme sounds like it will largely succeed on points 1 through 4. The ad revenue, once disclosed, will appear pitifully small for the first year or two. As long as trust is built gradually and testing provides insight, that revenue should pyramid up over time.
- Point 4 relates to Facebook -- in both senses. The network effect and addiction factor actually outweigh the fact that Facebook has been particularly brazen in doing wacky, unpredictable, privacy-invading things to its users. Facebook is very strong on point 1 and has point 2 covered also. Because its audience is very large, it can be cautious relating to points 3 and 5, monetizing below "potential," thus leaving long-term potential on the table. Huge win.
Some will question whether users will remain addicted to Twitter long term. Facebook is an entire social environment, and Twitter still feels like a "feature," a quick hit, despite a large user base on paper. That one hangs in the balance. Perhaps the litmus test for any would-be top-tier destination would be: are users choosing to download and keep their favorite mobile app related to that content, brand, function, or community, in the most accessible place on their mobile device? Will people get bored with them and stop? Will ads be easier to ignore on mobile devices? Will people look for versions of apps that allow them to ignore ads? (That's where point #3 really comes in.)
Change will be rapid, but based on these criteria, it appears that Twitter has the correct fundamentals and the right strategy in place for a long-term win. But if Facebook has a two-year head start here, you still have a nagging feeling that Twitter just needs to keep hitting certain user targets and to look reasonably dangerous revenue-wise, for the more realistic goal of selling the company to Google or Facebook.
Labels: online advertising, twitter
Tuesday, July 28, 2009
Nearly four months after their U.S. counterparts released the figures, the IAB Canada announces that Internet advertising revenues in 2008 grew at a blistering 29% pace, to $1.6 billion.
Even as a proportion of population and the size of the economy, this is below where it should be relative to the U.S. figure of $23.4 billion (up 10.6%), but the rapid growth is encouraging.
2009 has posed challenges in all sectors of the economy thus far, but both IAB announcements point to digital advertising as a mainstay of the economy, with their respective spokespersons highlighting the ongoing secular shift to online that drives growth even through a recession.
In the U.S., Search and Classified combined make up 59% of total digital spend with display placing second at 33%; in Canada the figure clocks in at 68% for Search and Classified combined, with display coming second at 31%. Curious about email? While still an important customer relationship medium, it's only at 2% of overall digital marketing spend as a prospecting method. In Canada that business is nearly dead with it making up less than 0.2%.
New studies need to be added to this mix to account for the full infrastructure costs of developing website technology, testing response, creating content, etc. How big is the investment "gap" in this area? Do companies plan to budget sufficiently in these areas? Those are open questions as larger companies make plans to redeploy the savings realized by reducing inefficient traditional ad spends.
Labels: iab, online advertising
Monday, November 24, 2008
While we certainly admit that paid search is best at converting high intent prospects into customers and leads, and it isn't always the greatest for other kinds of demand generation, the recent escalation in citations of studies -- like the Atlas study cited in a recent SEL article that shows that "93-95% of audience engagements with online advertising receive no credit at all when advertisers review campaign ROI" -- smacks of protesting too much.
Upon further, close study -- with a large panel for example -- you can show that more folks viewing a wide swath of display ads online are slightly more likely to purchase than the control group not exposed to those ad impressions. From here, we are supposed to "attribute" x% more ROI to the display ads.
But wait. These studies don't make much effort to ask additional tough causal (ahem, we are calling it attribution, so we are assuming causality, so...) questions, such as, did those exposed to the display ads get exposed to them because of any of the following?:
- They're a member of an enthusiast group that likes a certain topic, and so for that reason they are more likely to read the content on niche websites, so for reasons of being such an enthusiast in the first place, and being exposed to that content, at least as much if not more than being exposed to the display ads, they are in turn more likely to be found buying related items within the study period? Let's say I have warts, and go to a warts information site, that happens to run a few ads for "Warts Off." I happen to buy "Warts Off" in a drugstore based on some urgent need, and the colorful display case in the drugstore, that month. How much of that purchase intent, if anything, do I attribute to the online display ads? Who knows?
- They are exposed to a heavy blitz of advertising on: TV, radio, print, outdoor, in-store, direct mail, etc. For that reason, based on searches, serendipity, etc., they're more likely to be exposed to certain display ads as well. As a result of the overall media buy, that group does buy a product during the study period. But arguably, the online display ads being supposedly viewed during that period had relatively little impact on the purchase. Sometimes they play a significant role, sometimes they play very little role (spurious correlation only). Sorting out one case from the other isn't high on the agenda of those who sell digital display ads for a living, needless to say.
- Add to this second point, brand equity built up over centuries. Maybe some kinds of advertising lead to more purchase behavior, and others don't. And others lead to more online engagement generally, with display ads being part of that journey but not, perhaps, adding significantly to purchase causality. Perhaps the test group is more susceptible to the control group generally to this historic brand equity, and the increase in their viewing of relevant online display ads is caused by the impact of that historic brand equity on their online behavior overall.
Only one takeaway here, really: there's nothing wrong with giving a lot of credit to the channel that produces truly measurable ROI, and if you're prepared to go through contortions to "scientifically" prove "attributability" to other forms of advertising, be prepared to accept that the casuality debate may just continue to rage on, with everyone from product engineers, to infomercial producers, to "house article" writers, to word-of-mouth guerrillas, to everyone else involved with promoting a product or brand, to continue to clamor for their fair share of credit.
- Alexander Hamilton's face is on every $10 bill, but his brand isn't doing so hot. Thomas Jefferson, meanwhile, has a strong brand, and he's only on the 2, and there are hardly any of those in circulation. What is a fair CPM rate for either gentleman to pay for this type of exposure? This has nothing to do with the above points, but I thought I'd point it out. Food for thought.
Bottom line, if online display is so great and you can really prove that, the price will rise. If you can't, it won't. The judges of your science will be the ad buyers, in the end.
Labels: online advertising
Thursday, November 20, 2008
It's a bit surreal to juxtapose today's release from the IAB -- Internet Advertising Revenues for Q3 at $5.9 billion -- with Google's earnings release of last month, related to their Q3 earnings.
Since Google's revenues were $5.54 billion, you're apt to say: what the heck is going on here?
The IAB figures cover the U.S. only, of course.
But if you subtract Google's international revenues, their U.S.-only revenues for the quarter were all of $2.69 billion. So in the United States, this one company appears to be clocking about 46% of the revenues for the whole digital advertising sector. Twiddle a few knobs, and wait a quarter, and Google will be at around 50%. Among other things, you would imagine that this makes further acquisitions dicey for Google. One company moving from, say, 50% to 58% of share for an entire sector of such importance might make certain regulators nervous. (Not saying it's right, but that's sort of how the world works.)
Whether that's proof that Google is incredibly big, or the rest of the sector remains embarrassingly small given the gallons of ink devoted to it, I'll leave to your discretion. A third possibility is that display and other forms of online advertising have yet to reach the peak of efficiency achieved by Google's click auction.
Labels: iab, online advertising
Tuesday, September 18, 2007
Many Web 2.0 businesses fail to grasp the fundamental guiding principle of media and publishing over the past century: you're delivering an audience of "buyers" to paying advertisers. So says Steve Rubel in a thought-provoking post "Why Some Web 2.0 Sites Will Never Attract Big Ad Dollars," arguing: "Quantifying eyeballs is not the answer." Brilliant. I agree with Rubel wholeheartedly.
Naturally, some commenters were given to disagree with him, on the grounds that the web is not a direct response medium and that many of the blogs, social networks, etc. are hotbeds of value for advertisers (the "just trust us!" economics of painting certain online properties as "hot places" without any real proof).
A great way to prove whether your inventory is valueless or valuable to an advertiser is to put it to a harsher market test. Let's just say that there are some valuable online properties that never needed to persuade anyone they were valuable - not even advertisers. They made money because people really kept returning to these sites, and frequently enough, clicked on links while engaging with the content, and really did something, like booked a holiday. The accompanying referral revenues were 100% proof of value.
[Does that mean I think every advertising transaction should be CPA-based? No, of course not. Advertisers can see value without needing to complete transactions in that manner. But focusing on the direct referral model is a great way to simplify the debate: what if you had to give up all display advertising that had a CPM-based "brand" rationale, to earn all of your money from clicks that led to someone buying something from your partner?]
Not long ago, a great friend and educator who has been lucky enough to own a piece of a major dot com vertical site in the travel industry gave a private whiteboard talk roughly titled "Why TripAdvisor got bought for $200 million." In this informal chat, lines were drawn, consumer intent was graphed, and the fact that consumers at a point very close to purchase were using this site was clearly shown to translate into value to advertisers. The case was made that even without advertisers, the referral links to bookings engines were making money, and pegging a certain value on all that traffic. TripAdvisor, on the strength of this alone, was no doubt quite profitable. The referral partners (IAC), in the end, wanted to own TripAdvisor directly so they bought the company.
This example and quite a few others bear out what Rubel is saying. At the other end of the spectrum, there are huge swaths of online activity with zero commercial intent and thus, nearly no value. You can try to persuade someone of that value with fancy brand talk and WOM magic, but as the advertising market gets closer to being a true exchange, I think the low value of some of this online inventory will only become more evident.
Case in point: Google's contextual matching tech recently thought it had accurately seen a fit between a friend's (product oriented in a high tech field) keywords, his 40 cent bids on content targeting, and the published content on a huge number of pages on one of the major social networking engines (let's just say it was one of Bebo, Facebook, or MySpace). The traffic suddenly came fast and furious. 50,000 clicks and about $8,500 later, no one had bought a thing (zero conversions). 50,000 clicks on other swaths of inventory in that same account would usually bring about $30,000-50,000 in revenue. ($8,500 worth of clicks might have brought in $15,000 in initial revenue.) That's a very long way from the $0 achieved on that inappropriate inventory! Yes, arguably that inventory would have been appropriate for some advertiser. But I'm going to say: not very often, and not worth very much.
So often publishers see things only from their own distorted perspective. Look how big, respected, and impressive we are! Advertisers care a lot more about... you guessed it, direct response. Ad industry people selling things other than direct response often try to change the subject, and often there is a great case to be made for public relations impacts and other lifts that are harder to measure than direct response. But there's this thing about direct response... you can prove it. ROI-positive campaigns will end the "debate" about value and turn it into an ongoing ad buy.
There is some value to everything, but I'm looking forward to the furthering of models that give quality publishers a fair CPM rate for their commercially-relevant properties. Unfortunately that will also have to mean the poor quality inventory will continue to be classed as "remnant" until proven otherwise. Hopefully, the upcoming online ad downturn will actually be favorable to the proven properties who have a track record of relevance not just to readers and users, but to advertisers.
Labels: online advertising
Tuesday, May 01, 2007
DoubleClick may be working on a "NASDAQ-like exchange" for online display ads, but today ContextWeb has opened their ADSDAQ platform live, discussed here recently.
We're still trying to digest the relative impact of the DoubleClick (Google) and Right Media (Yahoo acquisitions), and to piece together their current and future impacts. There is already some speculation out there that Right Media specializes mainly in junk social media ad inventory -- if so, that's something for Yahoo to fix post-acquisition, or perhaps it's the part of the inventory pie that was available, so not to be sneezed at, especially not by the producers of large-scale junk social media ad inventory. (I do worry when you're using phrases like "non-premium inventory". In the past decade or so, non-premium online ad inventory has run the gamut from banners in less-visible positions, to forced page views. In other words, from near-worthless to fraudulent.)
ContextWeb seeks to broker both high-end and remnant inventory more efficiently by allowing publishers to communicate their ask price. ContextWeb is actively fighting the industry tendency to relegate publishers to "remnant" status by default.
The role of an independent growth player in the marketplace is a real wildcard. We know that product quality hasn't been the main reason Google won the PPC wars (reach was), but it was a real catalyst for pulling away from the laggards. ContextWeb's success will depend on the coolness of its platform (I'm confident of that) and the size of the marketplace it's able to build through business development and self-serve publisher signups.
Owning the premier properties seems to be the most favorable place to be here, given the emergence of rough-and-tumble competition among robust middlemen. Yahoo would really love to acquire Facebook, wouldn't they. So Right Media sounds like a consolation prize.
Labels: contextweb, online advertising
Tuesday, March 20, 2007
I was about to link to this site because they had a relevant article to my next post. But their article was nearly unreadable because of all the monetization around it. Hey, it's nice to sell ads, but...
So now they're officially in the usability hall of shame.
To review, they've got:
Is this supposed to be a joke? Too bad I gave up on the idea of reading the article here.
- A banner at the top
- A tile, top left
- Too much nav down the left
- Another top banner
- A "developer marketplace" text link unit
- A big square bugger
- A right-side skyscraper
- IntelliTXT underlining of keywords in content
- Another tile (lower left)
- White paper marketplace
- IBM stuff
- A single little text link ad for LocalLaunch
- Two extraneous widgets
- A large square AdSense unit
- A conventional AdSense unit
That's fifteen separate monetization pieces on a single page of content, an article that is unnecessarily chopped up into four parts to generate more "pageviews." Can you top it? Post examples, if so.
Labels: online advertising, usability
Friday, March 16, 2007
(via Searchviews) comScore begins touting a new metric, "visits," that allows a second visit and a third to be counted from a single user if that visit is more than 30 minutes after the previous one.
OK, well first of all, the concept of a visit is far from new. But let's run with it any way.
On the "pro" side, it's going to be vitally important to look at new measures of attention online. Time spent, and yes, visits, are key metrics because of the so-called death of the page view. New presentation methods such as AJAX will mean it's harder to measure pageviews (or impressions), and that makes it hard to fairly price advertising. I'd add that this is a good development in that content sites sometimes or often chopped up their articles unduly, or otherwise engineered a navigation model that actually produced more pageviews per visit. Should that be considered "more ad inventory" or not? Clearly we have always been dealing in nebulous concepts of user attention and some online advertising is well overpriced and some underpriced.
On the "con" side, I'll refer you to the perennial problem of sites gaming their traffic; or again, simply engineering more of what they have for sale. Valleywag's been relentless in poking holes in those who use small tricks to pump up the volume of what they're selling. The concept of visits is going to be unduly exploited by some sites, inevitably.
In the end, happy advertisers will be happy advertisers, and overpriced advertising will disappoint. But we do a disservice to clients if we don't continue to probe these various metrics for, shall we say, the "gaming opportunities" they might provide for some publishers.
Is "visits" better than uninformative stats like "reach"? Absolutely. Publishers and agencies have no business selling "reach." I'd rather see us be looking at a multifaceted stats package for pricing ads, like a quarterback rating (comprised of pageviews, time spent, heat mapping, and a list of various other stats). Or that could be considered at least a method of providing a third-party "scorecard" for how users work with a given website; of what type of audience and attention you're dealing with. You hear a lot of tough talk about third party advertising "audits," but that misses the point. You're not selling something that is ever 100% quantifiable, as the death of the page view shows. So as an industry I'd love to see us innovate, use scorecards, and explain the value of targeting.
Labels: comScore, cpm, impressions, online advertising, visits
View Posts by Category