Monday, January 11, 2010
Given the prevalence of studies that prove a certain lift in sales from costly display ad campaigns -- often showing indirect impacts of sales increases in populations "exposed to" the ads -- it's refreshing to have someone report an instance where the campaign had zero effect.
I've always been wary of the flaws in these studies. In general, you'd like to know what the overall campaign cost, at least. But in addition to that, being "exposed to the ads" to me still means you're also "exposed to the websites" and "exposed to certain communities" (or rather than being exposed to them, your likelihood of seeing these ads is driven by your membership and interests, which you had in the first place). The causal weight of the ads is still in question, if content and community are potential causes of sales patterns as well.
In this SF Chronicle piece
about Yahoo by reporter James Temple, we hear of a national retailer joining forces with Yahoo to measure sales lifts associated with a major display ad campaign. The good news: sales lifts came in at 5%, 93% of that coming in-store. The bad news: the sales lift figure for buyers under 40 was... zero.
If you're going to report the good, it's only fair to point out the bad. At least the ads didn't lead to a drop in sales.
Labels: display ads
Friday, November 06, 2009
It may seem like a small point, but "pay-per-click" was the nickname given to paid search advertising back when it started out, but it only describes a pricing method, not the nature of the media or what we seek from it. (In the proto-days of that same technology, "paid keywords" or "buying keywords" was another way of describing it.)
I was reminded of this whole mental muddle today reading the headline from an email solicitation, something about "getting more PPC without using Google".
But that whole line is kind of old hat. It's the come-on that opportunistic, non-search-based ad platform companies used to sell their crummy, remnant, and sometimes fraudulent contextual text ad inventory. Sometimes it couldn't have even been described that neatly. It was traffic, and you paid for the clicks, and potentially you used keywords to guide the system towards certain publishers, but that was about it. You might as well have paid the effective CPM rate, as bid on clicks. Didn't matter.
That's why I always advocated paid search as the term of choice for people who really wanted to go after clicks from ordered results placed near search engine results, but it scarcely matters what I advocate! -- people will use all kinds of terms.
SEM is another term that arose. Agenda-setters in the business tried to remind everyone that paid search (or "PPC") is one sub-type of SEM (search engine marketing), and SEO (on the "organic side") is another sub-type. But the fight to make SEM exclusively the global term, and not to be used as synonymous with PPC or paid search, was lost. SEM is often used interchangeably with PPC or paid search.
No matter. With all of these nomenclature battles being unwinnable, we should turn our attention to the whole reason "PPC" was so attractive as a pricing mechanism. It's because it represented a happy medium between CPM (paying only for impressions -- "cost per thousand impressions") and CPA ("cost per acquisition" -- paying for lead conversions or even revenue-generating sales conversions).
You can draw up equivalents across these mechanisms, and measure or express them all for your keyword (paid search) campaigns. So the click isn't anything special. It can be expressed in its CPM equivalent and you can and should also be measuring ROI, ROAS, or CPA.
Indeed, according to some scholars [see "Greedy Bidding Strategies for Keyword Auctions"], the most rational strategy for bidding in a digital media auction would take you straight to CPA or revenue if that was possible. If you could bid directly on the customer acquisition or revenue, you would. (And in fact, that's what some forms of bid management automation attempt to do, at one or two removes. And it's what manual campaign management also attempts to do, painstakingly.)
But step back further. Are search, keywords, or clicks inherently special? Why the drive to distinguish them from other forms of media? Is it for what they are, or what they represent?
It has to be the latter. They represent potentially the most extreme (and measurable) form of granular targeting and flexible bidding, of a certain type. This is reflected in the sky-high effective CPM rates for some keywords.
But that means that all of this distinguishing one type or another is done mostly for economic or practical reasons.
Search and keywords (and clicks) fall into the general category of auction-based digital media. Whether we're bidding on clicks, acquisitions, impressions, or other, the universe of digital media is amenable to similar tests. From a rational bidder's standpoint, there should be no inherently good or bad media, nothing inherently "creepy" or "wrong," nothing inherently above reproach either.
That's mostly true. It's not entirely true. (Dropping ad-laden anvils on prospects' vehicles is interruption media, and some companies would pay for it, but it's stupid and illegal.) But isn't it a good starting point for analysis?
"PPC" doesn't matter per se. So pitches like "now you can get 'PPC' from other channels than Google" shouldn't have any special weight. You shouldn't be looking too hard for that inventory if your economic criteria show it's not going to pay off for you. Nor should you have been ignoring it all along just because you thought that "PPC" or "Google" were special for some inherent reason.
Labels: contextual ads, digital media, display ads, google adwords, paid search, ppc
Friday, September 18, 2009
Google has finally launched DoubleClick Ad Exchange with full plans to integrate the system with the AdWords platform. Although a long-anticipated development, it's still a "wow" time for our industry.
Google joins a number of other true exchanges of note, companies that have pioneered the idea of a true "bid-ask" system on display ad inventory. The most notable of these are ContextWeb, having gone so far as to dub their system "ADSDAQ," and Yahoo-owned Right Media.
The designers of such systems are the best ones to provide detailed explanations of how ad exchanges differ from ad networks, but transparency of bid and ask prices is probably how to sum it up. If you're a publisher in an ad network like AdSense or the thousands of others that have graced the industry since the 1990's, there is no direct communication with ad buyers. The intermediary tells you what you're going to get on a CPM or CPC basis, either in advance, or after the fact. You don't get to enter into a direct transaction with the buyer, and you don't have any clear sense of what the intermediary's "markup" is. That might lead publishers and advertisers to want to make a lot of individual transactions with one another, to cut out the intermediaries.
And that leads to a patchwork quilt whereby inefficient individualized ad buying and selling is taking place on "premium" inventory, and networks are stereotyped as buyers and sellers of "remnant" inventory only.
Meaning: the state of display today combines two unattractive qualities: a lack of transparency and a lack of efficiency. That there are too many networks just adds to the inefficiency problem: it's an industry ripe for consolidation.
Jonathan Mendez, in one ("The Market Forces Killing Display Advertising") of a series of complex posts, seems to agree that this is the sort of problem facing the display advertising sector. He also, however, predicts that the new exchanges will only make the problem worse: they'll "drive down media costs even further and become a new haven for performance advertising at the expense of [publishers]." If so, I suppose that's great news for advertisers. But there could be a different long-term dynamic at work. Mendez's point seems sound in that he integrates it with a perspective that shows market forces from search advertising being rudely applied to the formerly fat and happy world of display. In short, direct response is what drives much of paid search, whereas something else entirely (brand integration) is supposed to drive display, like it does on radio and TV. If display is being forced to play in the direct response world, those who formerly profited from that channel are at a loss as to what to do next as their margins (and raison d'etre) fade. If I interpret even 10% of Mendez's message right, it's sobering food for thought, regardless of how many brand channel strategists Google layers on top of its direct response and search-savvy core.
Perhaps, though, it's simply far too early to make this call. I'm drafting a potential counterpoint to Mendez's analysis that roughly says: maybe the display ad market today is simply prehistoric and inappropriately organized as far as buying and selling dynamics goes; what if it's like Google AdWords in the pre-CPC, fixed-CPM, non-tested, non-measured, clueless-buyer era of 2001? What if the number of participants in a well-designed auction matters a lot to publisher revenues, and we simply need more? What if measurement and attribution right now are in the dark ages, and the introduction of assists will help? What if new technology (features) to rate publishers, classes of inventory, characteristics of content, got built into the system for either manual or automated use? But that could be another 1,500 words.
For now, a few more thoughts on where the DoubleClick ad exchange may take us.
Here are some principles to consider:
First, if a variety of networks can plug into the exchange and act as buyers, publishers may have a decent floor on their middling and remnant inventory. A "buyer" for a particular ad impression, then, could be any number of direct bidders participating in the DoubleClick exchange, or it could come in through outside networks who are also publishing in the DoubleClick exchange. That could mean a dual role for Google: (1) on one hand, hoping to create "the" platform for buying and selling ads, that improves the overall viability of the industry as it grows in size; (2) on the other, keeping the better outside networks in business, allowing them to buy and sell and even play around with arbitrage opportunities.
- Interoperability of networks: ecosystem sensitivity
- Continued migration of ad dollars away from inefficient media
- Applying the efficient auction principles of paid search to media buying as a whole
- Chicken-egg scale issues
- What kind of market is this? Will Google win with a capital W?
Second, the trend that won't appeal to traditional ad agencies and traditional publishers: dollars that can show clear ROI will be happier, so they'll be spend in this medium. The ongoing stampede continues.
Third - and this point is either scary or a forgone conclusion depending on your perspective - these principles will apply to all media someday: billboards, television, radio, product placement, etc. If you're a company like Google or Microsoft, you're thinking about organizing a platform to run that. Google did a pretty embarrassing job of doing this the first time around (radio, newspapers), which just proves you don't just snap your fingers and accomplish something like this. It's very early days.
Fourth, it looks like the early exchanges were sort of unsatisfying in the sense that their scale was understandably limited by how many participants were on either side of the transaction. If the buy isn't big enough, it's not worth the time to monitor your involvement with the platform. Signing up "sellers" (publishers) is a major prerequisite to making this work. And the pitch to them has to say something like: "we have millions of advertisers eagerly logging into our platform every day". Pretty much only Google (and maybe one or two other companies...maybe) can say that. This is a game-changer, potentially.
Point 4.5 is simply along the same lines, but it's probably important not to sidestep this issue. You don't draw up "market maker" logic in the abstract and then from there find great success. If the results to advertisers or publishers are lukewarm, they won't hide that fact, they'll simply stop using you. And unlike traditional ad networks and traditional media buying, your whole principle is that the exchange technology itself drives liquidity and sets prices. If that isn't working, you can't fall back on the sales team to grease the wheels; that would be incoherent. To date, some existing exchanges have suffered from the critical mass problems. Others have meted out condescending treatment to ad buyers, reverting to salespeople who promise to make a custom buy on the system, or work the system for you, as long as you commit to a certain budget. Hey guys, if you're trying to prevent me from logging in directly, then you're a mutual fund salesman exacting a fee; same old, same old.
So, point 5: is this a market like the one Amazon took by storm - a winner-take-all market where Google will enjoy market dominance? Or is it a market where first and second movers continue to do well (like the browser market, where Google only has 3% share, or the -opedia market, where the Google upstart product knol has nowhere the brand adoption of the original, Wikipedia)? How will Yahoo and Microsoft respond? For the record, I caught up with Jay Sears, EVP of ContextWeb, and he said: "We welcome Google to the ad exchange business. It's a terrific market validation to now have the two top exchanges in the market, including one that is independent."
We'll see how things shake out.
Something to think about: interoperable networks -- those less evolved creatures -- may do better in this ecosystem than competing exchanges. A clash of platforms is a true clash of superpowers; non-superpowers die. At most 2-3 leading platforms will win. Meanwhile, "app" creators (sub-players in the system) that work well within the leading platform are non-threatening to the leading platform makers, and don't need critical mass to be profitable within that environment.
In any case, to conclude by explaining what I meant by the title of this post: the display ad business is in a frustrating state, as every company developing products and services to serve that market seems to tell you in their booth pitch. The problem is that virtually none of these companies solve the problem; most make it worse, or sell you a futuristic solution adorned with "truthy" FAQ's, and then revert to old-school methods. In 2009, it's simply a myth that there is any satisfactory display ad system that is built to scale with a scaled-up marketplace of buyers and sellers. If that scale is reached - and I don't think it will be before 2012 - we may be in a quantitatively new ballpark.
Labels: contextual ads, display ads, doubleclick, google
View Posts by Category