Being John Malkovich, media planner edition

Alright, I got a budget from a client.  We have a job to do - it’s a big client, a big brand. Of course they want people to buy their products but we also know that it’s not going to happen on impulse. And it not necessarily going to happen online. So it can’t be all DR, it has to be somewhere in-between.

Breathe in, breathe out. SEM gets a piece, it’s a no-brainer. Lower funnel, gotta work it. Volume won’t be as high as I would want. That’s just common sense. And I read studies that show how display adds lift in search, maybe 155%. Google published a study like that recently, too. Great - let’s do display too. Hopefully there will be good offers in place and the creative is good. Alright.

Display. OK.  Breathe in, breathe out.

Ad networks, DSPs, premium direct publishers.

How do we optimize? Is it on clicks?  Nah, there is the “natural born clickers” problem. Can’t really do it on clicks. Well, I could, but I wouldn’t know how to justify to the client if he has read that study. If I optimize on clicks I am optimizing only for a small group of people who click - the client’s customers are much more diverse than that.

Well, there is also the viewthrough conversions or even visits. Google says it’s good - there must be at least some truth to it. We could do that to pad the results of the campaign.

For brand awareness lift we could use Dynamic Logic or something similar - it will always show a wee lift, we could extrapolate and say that for the entire campaign can expect the same lift as on premium sites where DL was used. Cool.

Speaking of sites, it’s much cheaper to go with any number of the hundreds of ad networks out there, plus, the vultures they are, they will optimize for me and schmooze me out for lunches. But I’ve read that online advertising is more effective when it appears on a site that you know and that you trust. So premium would be better, but if performance is 5 times better, and the price is 10 higher, it doesn’t make sense. Or does it? Plus I’d have to negotiate prices before coming up with the actual ratios.  Would have to call a bunch of them and make deals. That’s annoying. 

And what about frequency? The more ad nets and publishers I use, the higher my average frequency could be if there is duplication between those sites.  And there is duplication between sites, of course. People hate high frequency. It would better to just do one buy over a DSP and manage frequency through one channel.  Whatever - that’s the price we gotta pay, can’t put it all in DSPs for now. It’s going to take a while to figure out which DSPs to use - and since ad networks buy over exchanges maybe I could just throw all of my budgets through a respectable ad network.

Uh, but we also need to show the client that we’re an innovative media agency, have to create some strategic control.

Data.

We could buy data, accumulate data, work our data. We’re still using a DSP like other agencies, and we got a multi-year deal for a third-party ad server from GOOG/Microsoft, so it doesn’t really matter, of course - we’re not creating much strategic control, but it’s a message that sticks these days.  And where is this data we’re buying coming from? Publishers who sell data don’t want to be upfront about it, so there is a lot of trust involved with data exchanges. Kind of like with “behavioral advertising”, whatever that means these days, it’s just that the use of the data is now my own responsibility instead of someone else’s.

And what about retargeting? Have to drop some pixels on the advertiser site.  Ugh.

Almost forgot! What about auditing! Don’t want to be ripped off by any of the ad nets and direct publishers that throw junk inventory my way. Below the fold or above the fold? Does the creative even go through the viewport? And does the fold matter? Designers say it doesn’t! But it would make sense to at least track when the creative crosses the viewport. Oh but it will only be for placements where we don’t go through iframes and maybe some publishers or ad nets won’t let me do it.  Will have to try anyway.

DING! New email in my inbox. Looks like it’s the creative for the campaign!

This creative makes no sense. It’s a set of the ugliest 300×250’s I’ve ever seen (or since the last campaign, at least). The animation is way too long before the message becomes comprehensible. The call to action is terrible. Don’t understand what the offer is. But this went through a very long round of client approvals and there is no way to go back.

…I guess I’ll have to make it work somehow.

The Crowded Buy Side

There’s been a lot of buzz about Terry Kawaja’s keynote at the recent IAB NE event. AdExchanger has posted the presentation and the video, “A few good DSPs” in case you’ve missed it.

The most interesting slide in that deck is the ecosystem map that Terry has worked out. We already knew how bad it is, of course. There are a lot of players in the marketplace. It’s a complex value chain where margins will be inevitably get squeezed. Too many players going for the same pie. Etc.

What’s striking in this map, however, is the imbalance in the number of companies being started on the advertiser side versus publisher side.

So why is this? Why is it that most ad technology companies are being started on the buy side? Is it because the greatest value to be created for advertisers resides in how media is bought and sold, and how precise the targeting is? Surely all these companies can’t all believe that given any 300×250 or 728×90 the client throws at them, they can make sense of it and create value for the brand, simply by adding targeting options and making inventory cheaper to buy?

The main reason is very simple. You can change the actual terms and start talking about DSPs, SSPs and whatnot, but at the core of all this is still good old ad serving. And when it comes to ad serving, the barrier to entry is much, much higher to work with publishers. A provider of buy-side technology can approach any client or agency, ask for a small budget to try out their “buying platform” (whether or not it even exists), and they can make it successful on a small scale with relative ease. The agency can include this new placement as an additional line item in an already long media plan.

If something goes down, it’s really not a big deal. In the case of RTB you simply won’t serve and no one will incur any cost. In the case of direct third-party ad delivery, you can make appropriate arrangements with publishers to take you out of rotation.  Failing is really not that bad. The big guys such as Yahoo! certify you and make you sign SLA agreements but for the most part, it’s not a problem.

Compare this to working with publishers. Even taking into consideration how bad the legacy platforms they run are, convincing a publisher to change first-party ad serving technology is incredibly difficult.  It’s mission-critical stuff. There are a lot of people to train.  You simply cannot go down. If you do, money is being lost, literally. On the technology side there is a huge complexity of pacing and scheduling campaigns given their delivery objectives. As a publisher, you need to deliver on your commitment of volume. Data aggregation is also crucial and is very complex to manage on a large scale (see this post by Mike on Ads for more on this).

On the upside, despite the complexity, working directly with publishers really allows to create value for advertisers. It brings you closer to formats and site layouts, user profiles and data. It brings you closer to participating in the creation of new ad products that advertisers and agencies actually buy. It allows you experiment with them and showcase your value.

With AdGear we took the approach of starting with first-party ad technology, spending the time it took to work out the hard stuff. Fortunately, in this new RTB world first-party technology is also a huge asset for advertisers and agencies as well, since you essentially end up scheduling and pacing campaigns over a huge pool of inventory (albeit with more targeting and filtering).

No matter the number of players in that ecosystem map, chances are that at the end of the ad serving chain, it’s still the good old Atlas, DFP or OAS making the scheduling decisions and stopping delivery once the volume objective has been accomplished.  These legacy platforms are still the backbone of what we consider to be the new value chain.  The good news it that there is an increasing number of companies that are taking a stab at first party ad tech, too.  And that is a good thing for the quality of advertising in the medium.

Stupid RTB Tricks

At BLOOM we’ve been integrating real-time bidding functionality and have been running campaigns through AdGear since November 20o9. We’re plugged in to the DoubleClick AdX 2.0 and although our volume there is relatively small compared to the “traditional” third party delivery, it’s been a lot of fun.

To anyone dealing with the sheer scale enabled by RTB, it ends up being a boundless source of ideas for very questionable practices. This, in turn, raises the issues of privacy and data ownership. There has been a lot of coverage of RTB, and of course AdExchanger has done an amazing job in covering the new generation of ad companies (including us!). But let’s talk about the dark side of RTB.  The Stupid RTB Tricks. Media hacking.

1) Browser history retargeting.  Imagine that you have a two-sided display campaign. On one end, you run on sites or networks bought direct, with huge reach. Everytime you serve an impression, you check user’s browser history (see a demo of the CSS trick here). What you check for are your advertisers’ competitors’ sites.  In the second part of the campaign, you retarget those who have visisted your competitors’ site with a sweet call to action.

2) Über-browser history retargeting. Adding on top of the first trick — create a segment for those who have visited more than one of your competitors’ sites. This might indicate that the person is in a serious shopping mode. Since browser history CSS trick works based on actual URLs and not just top-level domains, you could go really deep inspecting which products on your competitors’ sites were looked at. Targeting this segment through RTB is a no-brainer — bid like you’ve never bid before.

3) Using premium media as audience qualifiers. Buy a direct campaign with a premium publisher that reaches a very valuable audience. Say, WSJ. Run the campaign with frequency cap of 1. Reuse this data to serve to the same exact audience across the RTB universe. Do so for more than one advertiser, compensating the original advertiser whose budget went to WSJ.

Of course, there are many more opportunities of this sort.  And although there is nothing really new here, having the possibility of doing it “on tap” without necessarily having a huge scale in the retargeting segments is definitely new. You don’t need to setup anything special with ad networks and run 10 witnessing tags from 10 of them, don’t need to commit ad budgets for retargeting without knowing what the final reach will be, etc.

I’d love to see other people post more examples of media hacking (both in the good and bad sense of it).

AdGear preview

So we’ve been working on our ad management platform for a while now, here is a wee preview of it. Check out adgear.com too!

Microsoft and OpenX?

Online advertising start-up OpenX and Microsoft Corp (MSFT.O) announced a partnership to promote each other’s Internet technologies to their respective customers.The deal will significantly expand the distribution of OpenX’s technology for serving ads on websites, as the company seeks to boost its roster of large Web publishers, said CEO Tim Cadogan.

OpenX will also make it easier for its Web publisher customers to use Microsoft technology that analyzes the content of a Web page and matches relevant ads to the page.

Microsoft and OpenX forge Web ad deal | Markets | Markets News | Reuters

What is never clear to me is how many of OpenX’s 50 million publishers are relevant. What is the proportion of these that runs local installations versus hosted/market ones? Or does the local installation now include a market component?

Kind of an odd announcement for Microsoft, isn’t it?

Quote from: Stackoverflow, Advertising and the Ethics of a Free Lunch ~ C for Coding

If the Internet has taught us nothing else, it has taught us that:

1. Advertising pays for otherwise free services;

2. People don’t like advertising; and

3. Advertising works. These conflicting forces always cause consternation and Stackoverflow is by no means immune.

Stackoverflow, Advertising and the Ethics of a Free Lunch ~ C for Coding

comScore and Starcom USA Release Updated “Natural Born Clickers” Study Showing 50 Percent Drop in Number of U.S. Internet Users Who Click on Display Ads - comScore, Inc

“The act of clicking on a display ad is experiencing rapid attrition in the current digital marketplace,” said Linda Anderson, comScore VP of marketing solutions and author of the study. “Today, marketers who attempt to optimize their advertising campaigns solely around the click are assigning no value to the 84 percent of Internet users who don’t click on an ad. That’s precisely the wrong thing to do, because other comScore research has shown that non-clicked ads can also have a significant impact. As a result, savvy marketers are moving to an evaluation of the impact that all ad impressions – whether clicked or not – have on consumer behavior, mirroring the manner in which traditional advertising has been measured for decades using reach and frequency metrics.”

From comscore

Nothing really new there. It would be interesting to consider the most active advertisers in 2007 vs. 2009. From a quick look at the TNS data in March 2007 and March 2009 it seems that the financial industry is spending more in 2009, but could also be that the proverbial long tail of advertisers is also getting longer. So, more small advertisers generate impressions no one cares about. Interesting stuff but the idea of “clickers” and “non-clickers” is kinda useless. Can we generally separate people interested in products/services/things and others who aren’t? Would that work in any other medium?

Intent generation versus intent harvesting

A commenter on my blog the other day (Tim Ogilvie) mentioned a distinction that I found really interesting between intent generation and intent harvesting. This distinction is critical for understanding how internet advertising works and why it is broken. It also helps explain why sites like the newspapers, blogs, and social networks are getting unfairly low advertising revenues.

Why content sites are getting ripped off — cdixon.org – chris dixon’s blog

Clever way to express the inequality. The whole chain is more complicated since search engines also provide content sites with traffic but whatevz.

And I guess I can see how Hunch fits into this (since Chris is a co-founder)

Ad Network Differentiating Factors

Data is also improving. But because it is also becoming more of a commodity, the real question will be whether this data can be proprietary.

Where will the next ad network breakthrough come from? « Lightspeed Venture Partners Blog

The question asked by Jeremy Liew is a great one and the post is worth reading. I’d have to add that one of the things ad networks overlook is actual metadata on the sites, placements and pages that are part of the network. 

This data is hard to collect, it’s hard to categorize and it’s even harder to make use of given the legacy ad servers they all use (hello DoubleClick), but this would allow networks to create other, new vectors of targeting. 

Imagine the ability to target a campaign based on how many ads there are on the page? Or whether the placement is above or below the fold? Or based on how prominent the placement is on the page versus content? How much time on average people spend on that page? All these criteria warrant a premium that advertisers would gladly pay. 

Since inventory is abundant, you have to separate the good from the bad and treat is accordingly on the sell side. Ad exchanges don’t solve that problem, but ad networks are well positioned to differentiate themselves this way. The same goes for publishers too, of course. 

Is the DoubleClick Exchange a game-changer?

Yesterday Google announced the launch of DoubleClick exchange, sending ripples down the blogosphere and beyond. A good list of discussions is available at TechMeme, one of the best ones being Andrew Goodman’s at Traffick.

Reading the blog post alone can be confusing. The platform is positioned as DoubleClick Exchange, specifically, and not Google Exchange. DoubleClick is the enterprise brand that completely dominates the ad serving market and has reach over 80% of large publishers and probably just as much, if not more, large brand advertisers. Why, then, does Google talk about integrating this platform in Google AdWords, the openly accessible ad platform where you can open an account with 10$, and not DoubleClick for Advertisers which agencies use? All while maintaining that this initiative “helps large online publishers on one side; and ad networks and agency networks on the other”?

Large publishers, after having flirted with ad networks and ad exchanges such as Yahoo’s Right Media, are notoriously cautious of anything that can commoditize their inventory. The big concern, of course, is the notorious pork bellies problem which occurs when advertisers cherry-pick inventory and throw away impressions that don’t interest them. By far, that is the publishers’ highest priority given the rise in popularity of real-time bidding mechanics and the loss of ad budget commitments that comes with it. It just happens that this is also the exact value proposition of ad exchanges to advertisers and agencies.

There are two ways to make this viable for an exchange: 1) Try to fill all the inventory holes created by real-time bidding; and 2) Deal mostly with remnant inventory, creating little value compared to competitors.

The market is not nearly mature enough to make the first option happen anytime soon which will reduce the DoubleClick Exchange to a glorified AdSense for display, focusing mostly on the long tail and not making much of a splash in the brand advertising world.

There are some moves which may make this more interesting for advertisers, such as deep integration of the exchange in DoubleClick for Advertisers,  but Google may have to risk some of the handsome revenue they make on ad serving from the DoubleClick products somewhere along the line. Imagine a scenario where the publisher uses DFP, the advertiser uses DFA, and a third party network uses the exchange. That just seems like too much dipping.

Generally speaking, ad exchanges will become an important component in the creation of a global ad ecosystem where hyper-targeting is universally possible across multiple properties. A big part of this problem remains the legacy ad serving infrastructure used by large publishers and for which DoubleClick is largely responsible.  It really feels like it’s finally time for the next generation of ad serving platforms — a generation that plugs into exchanges and data sources alike in order to create actual competitive advantages for publishers and for advertisers.  From the technology standpoint, publishers don’t have to be part of marketplaces or put their inventory on the block to enable third-party targeting and truly leverage data.

It’s great to see at least some development in the market that is, as Andrew Goodman says, “in a frustrating state.” Too many are talking way too much science fiction and not doing enough problem solving.  We don’t need to do easier display advertising (as Google positions the exchange). We need to derive and show more value, which today means more integration on data and formats, and that definitely does not mean equating it to direct response.

Let’s concentrate on that, shall we?