Death Of The Impression/Rise Of The Data Economy

Recently there has been buzz around the rise of advertising exchanges, which bring media inventory into a liquid, dynamic environment driven by automation and technology. The exchanges come in many flavors, but they all move in the direction of allowing wide swaths of media impressions to be bought in a computerized manner. Google, Microsoft, and Yahoo are all in the game, with ever-increasing amounts of inventory becoming available in this environment.

In one sense this movement is a natural extension of the race to build the bigger and better network that can offer the one-stop shop to agencies and clients. Aggregate enough impressions, guarantee reach, sprinkle in some targeting solutions and you have a winner that can appeal to a wide range of clients. It is a simple enough model and an easy extension of the traditional media buy. But beneath the surface of this move is a far greater shift that will shake up and unsettle marketing as we know it.

It is about data -- data in ways we have never before fathomed. The future of advertising is not about social, not about viral videos, not about mobile, not about any new medium or any new ad unit -- but about data. Those who know what to do with this will be the new kingmakers, the new rulers of Madison Avenue -- or the creators of a new Avenue of media.

Why is this so? Because the impression by itself is becoming worthless.

In the past media was a relatively constrained commodity with a finite amount of time or attention that could be rented. There were only so many publications, so many channels, so many avenues to reach the audience you coveted.

This is no longer true. We have reached terminal velocity, the exponential, the moment when it is clear that the number of media vehicles and opportunities has far exceeded the ability for any marketer to manage without the use of technology. Welcome to the beginning of a seemingly infinite media world.

Some perspective: there are now over a trillion known Web pages, more than 500,000 Facebook apps, 140,000 iPhone apps, billions of videos on Youtube -- and more have been created since you started to read this sentence. The depth and breadth of content is mind-boggling, and we have only just begun. All trends point to an accelerated world of content, as ever-larger percentages of the global population come online armed with easy and powerful tools for capturing, creating, spreading, and consuming digital assets.

The standard way to deal with this phenomenon has been to work with networks that can aggregate this inventory into manageable bundles. However, this has its disadvantages. The fundamental problem is outsourcing the core understanding of what works for the business. The strategies, audience data, and all of the other factors that make a media buy successful are hidden to the marketer. Want to buy on a new network? It's time to start over again.

This is not the only problem. Buying from multiple networks? Chances are you are bidding against yourself because many of the networks are vying for the same inventory sources behind the scenes. Looking to find your customers via remarketing? Add yet another tag to the site and give another vendor your valuable customer data.

It gets worse. Most of the standard methodologies for buying an audience do not translate to this new world. When there were only a few channels it was pretty easy to get a good sense of the media with which a certain demographic resonated. When there are infinite channels, this approach doesn't scale. The person in the cube next to you might be an exact match on every normal demographic point and yet have a radically different media consumption pattern.

In this new world, the successful brands will be the ones that speak directly to you. These brands will be able to identify not that you are a likely traveler -- but that you are traveling to Atlanta next week. They will be able to tell not that you are likely to play games, but that you are a hardcore street fighter player with your own custom joystick. The brands that get you are the ones you will reward with your time, your attention, and your wallet.

The critical component that makes this new world work is data -- not simply general research data, but data about you. This goes far beyond just behavioral targeting, to your preferences, your interests, your actions -- all of the signals you send as you move through the grand stage of life. The revelation is that this new world is no longer the far-off land on the horizon -- we've hit the beach.

Right now the new world looks something like this: a global Wall Street of ads in which media trading firms will bid for your attention at every part of your life -- when you are at a computer, using your portable device, in front of a digital billboard, listening to your personal radio station, chatting with your social network, flying for business, playing your favorite game, and numerous other modes of interaction we could only dream before. A world where we can connect to the Net whenever, wherever, and however we want is a world that presents an infinite variety of means for an advertiser to engage and interact with us.

The first preview of how this looks can be found in Google. For not being a fan of advertising, Google has certainly done a good job at it. What makes Google different is that the ads are content themselves. The better targeted the ad is to the user's interest, the more often the user will click -- and the more money Google will make. Its entire ad system is built to encourage advertisers to tailor their offer as specifically as possible to what users are searching for. It may not be a perfect system, but it works.

Google also hints at what the future of publishing might look like. While Google is a search engine, it is also a publisher: one whose content is search engine results. In that sense, Google might be the most highly monetized publisher on the planet.

The lesson for publishers is that the more you know about your users, the better the experience you can provide for them -- both in content and ads. Already moves are being made in this direction from companies that aim to open up premium inventory on the exchanges with a range of targeting options and controls.

An open marketplace based on data will reward the best publishers for the attention and trust they have gained from quality users and enable advertisers to quickly and fairly price this attention -- and customize ads to fit. Publishers that wall themselves off will eventually find themselves irrelevant, as marketers buy their audience without them.

When we turn to the current aggregators of publishers -- the networks -- it is clear that disruption lies ahead. In one sense networks have been the first to use the impression and data exchanges to extend their reach and targeting capabilities. Many agencies and advertisers have been playing in this new economy without realizing what was happening in the background to make their media buy successful. Increasingly, however, marketers will access this inventory and data directly and cut out the high margins that the networks have enjoyed.

This shift will splinter the network world into a few different factions. Some networks will become more like agencies and partners for brands and use their optimization and campaign skills to grow a client's business. Vertical and specialist networks that are site-specific will still exist, along with networks for emerging media that has not yet been commoditized by the exchanges.

The losers in this new economy are the networks that have only existed for reach without a high degree of additional value and insight. When marketers seek ever-higher returns for their dollars, it will be difficult to justify the extensive arbitrage that has played out in the first version of the ad economy.

Just as networks will have to look more like agencies to survive, agencies will increasingly have to look more like networks through the use of custom technology, proprietary trading strategies, and a deeper understanding of how to leverage vast pools of liquid inventory against massive data sets. Instead of media buyers, there will be media traders whose jobs look something like a blend between a Wall Street number jockey and an Army field general. Premium negotiated placements and custom deals will always be part of a large media buy -- but the days in which most deals are done over drinks and a handshake will look like the Wall Street paper-slip days of yore.

There is the question of creative and the role of the message. Core brand positioning has an intrinsic value, but sophisticated marketers are already learning that it is not about one message, but the right message at the right time to the right person. The more data and insight marketers can understand, the greater the ability to customize the ad for that user at that moment -- yielding engaging experiences and high ROI.

This trend can already be seen in the e-commerce space, with giants like Amazon that have pioneered smart recommendation technology and ads that are targeted to your preferences and choices. As this technology filters down, ads will grow increasingly intelligent and persuasive.

Finally, there is the data itself. Direct marketers have known the value of a good list for years, but this is an entirely new playing field. Data will come in all shapes and sizes: from your standard demographic breaks to infinite categories of interests, activities, thoughts, locations, relationships, purchases, and any and all properties that can be abstracted into a database table. Data is the oil that fuels every part of this new ecosystem and the real challenge for the marketers of the future/now is how to understand, price, and act against it. Google, Amazon and the like have understood this for many years and are using their information hegemony to dominate existing markets and terrify new ones.

Numerous issues and practical realities remain to be solved, from ensuring adequate privacy controls to brand protection safeguards, but there can be no doubt that the shift has already begun. The only question is how you will adapt to survive and thrive in this new media economy before the gravel road becomes a quantum highway.

24 comments about "Death Of The Impression/Rise Of The Data Economy".
Check to receive email when comments are posted.
  1. Nelson Yuen from Stereotypical Mid Sized Services Corp., February 19, 2010 at 11:36 a.m.

    I couldn't have said it better myself. (Nor would I ever try.) Seriously, a MUST READ.

  2. Mark Walker, February 19, 2010 at 11:51 a.m.

    The impression sell worked for about 50 years in broadcast, but now we have options that were never-before-available to us as marketers. It is a sad death for the impression, but long live the responder!

    I market the largest off-airport parking facility in the world and tell ad reps almost every week that impressions are nice, but they don't park on my lot- I need responders and customers!

    Consequently, my branding budget has been reduced in favor of more measurable and response-based tactics. When the economy comes back- or when my competitors wake up- I will likely expand branding efforts. But for now, I want to try media that deliver results.

  3. Josh Grossman from Springpad, February 19, 2010 at 12:17 p.m.

    Whole heartedly agree. This is the most important line in the piece: it is not about one message, but the right message at the right time to the right person. This is the gold mine for marketers and data is at the core. One of the things we do at Springpad (, is not only help people save anything they want to remember, but actually treat everything saved as a "data object." From there, when users access the stuff they've saved, marketers can know that this "data" is something a user is explicitly interested at that moment in time - and that has great value. Wouldn't Perdue love to know that a person is looking at their saved chicken recipes for their dinner plans. The key is known intent versus implied intent on the part of the user.

  4. Michael Andrew from Mediasmith, February 19, 2010 at 1:36 p.m.

    Josh -- I think that is a great example of how user intent could be used to serve users ads they might be interested in. Services like yours could share the data and preferences your customers indicate with the content they have selected (with their clearance/consent). There are some mechanisms now where you could do that and the tools are growing richer by the day.

    Personally I'd much rather see ads tailored to my work and personal interests than the generic ads that are blasted across the internet (and real life!) for any and all to see. Better Data + Smarter Advertisers will hopefully lead to a future where everyone wins -- the consumer and the marketer.

  5. Michael Andrew from Mediasmith, February 19, 2010 at 1:43 p.m.

    Mark -- agree that measurement and response based tactics are a critical part of any campaign but I don't want to knock branding. Much of the data coming online actually helps advertisers to do *smarter* branding and reach the target that matters in an effective manner.

    As the data gets better and easier to handle we will be able to measure across the entire life cycle -- from the first discovery down to the bottom of the funnel where the user makes their purchase decision. We can do much of this now but the richness of the data that is being added for targeting will help make for a completely different experience.

  6. Brian Suthoff from Localytics, February 19, 2010 at 1:51 p.m.

    Great piece. We see this in spades in mobile, where the personal-ness and ubiquity of mobile contributes context and engagement beyond that of the fixed web.

    Mobile analytics solutions like ours ( are already being used by brands building iPhone apps to measure customer engagement, and by publishers to drive exactly the insights you mention. It's a very exciting new world!

  7. Domenico Tassone, February 19, 2010 at 2:33 p.m.

    Maybe that is not your intention but "Death of the Impression" sounds like hyperbole and throwing out the baby with the bathwater.

    Branding and response isn't always an either/or proposition.

  8. Dave Woodall from fiorano associates, February 19, 2010 at 3:10 p.m.

    This is a fantastic piece Michael but I feel like news of the impression's death is greatly exaggerated.

    The "key" still boils down to the bottom line. Yes, one-on-one messaging is the Holy Grail but it's not hard to envision a number of possible scenarios where the use of data that is too granular becomes counter-productive.

    In the hyper-targeting future, does 7UP never target a Coke or Pepsi drinker because they've never shown an interest in caffeinated sodas? Does Mercedes-Benz ignore those who've driven domestic brands their entire life? Does Royal Dutch Shell bail on "low-value" hybrid and EV owners? Advertising's job is to sell. Maybe not today, or tomorrow, or even this year but the key is to - at a minimum - have positioned the brand so it at least receives consideration; if you never ask for the sale you'll never get it. Among consumers with no brand preference or those seeking to switch brands, hyper-targeting will cause many intenders to be passed over.

    And then the question becomes: Are we trading efficiency for effectiveness? Are we converting a greater percentage of our hyper-targets at the expense of our secondary and tertiary targets? Is the result a net increase in profits?

    Obviously no one knows at this point but regardless, I think there will always be a place for impression-based buys. The only question is "How big a place?"

  9. Michael Andrew from Mediasmith, February 19, 2010 at 3:32 p.m.

    Dave & Domenico: On the death of the impression what I was trying to get across is that in a world of limitless inventory an impression by itself will have no value if not bundled with data.

    The quickest ones to use this data will be direct response advertisers but I actually think the long term value is in allowing advertisers to buy *audiences* to drive their branding and response goals. Right now most advertisers will buy content assuming they are reaching a particular audience -- but this might not be true or the composition might be completely different than they imagined.

    In the 7UP example what if they could specify that they wanted to show an ad to any male between the ages of 18-25 with certain income characteristics who is currently located within 1 mile of a gas station that sells 7UP while they are using their GPS enabled device -- and highlight that in the ad? Branding + Response at the same time.

    In the end all of this serves the aim of helping a Brand connect to the consumers that are a fit for their product through every stage of the buying process. Smart advertisers will learn when to show the general brand message and when to try to close the sale.

  10. Brad Boekestein from Wallrich Landi, February 19, 2010 at 3:45 p.m.

    Maybe one of my favorite articles of all time. I have been trying for months to understand the rapid shift in data reporting and metric analysis. With so many sites popping up and new social networking sites launching, it's becoming increasingly difficult to reach the demographic you want.

    I look forward to the day when someone comes up with a way to successfully reach each fragment of the market. Consumers and businesses alike will rejoice.

  11. John Grono from GAP Research, February 19, 2010 at 5:15 p.m.

    While I agree that at the pointy sales-end of the product sales chain data is going to be increasingly important, I have a couple of caveats, based on 33 years of linking marketing data to sales.

    First, to look at your travel analogy, that the brand that knows that I am travelling to Atlanta next week will win out. I am far from convinced that this is the case. There are/will be a plethora of products out there that offer that capability, but the one that will get my business will be the BRAND that I trust to get me there on time, safely and at a reasonable cost. If I am on a site with a dozen options, then the airline with the good safety record, good cabin service and check-in, that is not price-gouging will get my business of "Fly-By-Night" airlines that has tracked me and offered me a flight at price that is so ridiculously cheap that I begin to wonder whether they will have a crew or enough fuel on board.

    Second, with a trillion web pages all trying to monetise themselves to a population of 6 billion (of which around half are online), serving a mountain of impressions every day, no wonder the "impression" is in danger. And what do I see as the response - let's serve more ad impressions. Is it out of the realms of possibility that so many online ad impressions are served that no matter how targeted they are, that they all end up as wallpaper?

    As a fulfillment (read sales conversion) medium online is superb - I do so much of my business there every day. But it strikes me that despite the mountain of impressions I am served daily I am not getting a sense of any brands being built by those impressions that either 'yell at me' with screen takeovers or are just an annoyance to my real purpose of being on that webpage.

  12. Kevin Horne from Verizon, February 19, 2010 at 5:38 p.m.

    Can't argue with any of it but I would add these items to consider:

    1) Many, many marketers are much farther along this path than you imply - we'll hopefully see loads of case studies popping up middle to end of this year

    2) I believe we are already going up the other side of the pendulum now with a "data only/data is magic" emphasis. Altho i get your point that supply is now limitless, not all of it is important (no one is advertising on my blog, for example).

    3) Related to that, I think we're in danger of understanding "why" of the data. Maybe it will become meaningless once we have enough bits and bytes on everyone's every move, but i think we still need some good insights into the "why" of customer behavior...specific to segments, product categories, etc.
    I would point you to a blog post i just wrote about that, but that would be overly gratuitous so I'll leave it at that.

    my three cents...

  13. Kevin Horne from Verizon, February 19, 2010 at 5:39 p.m.

    oops, point #3 should read

    "...I think we're in danger of NOT understanding THE "why" of the data..."

  14. Michael Andrew from Mediasmith, February 19, 2010 at 5:44 p.m.

    John -- I'd agree with you as it stands now that online has not served as the best vehicle for branding and that Brand is highly important in influencing that last purchase decision.

    But if we look 5-10-20 years in the future where suddenly everything is online AND we have good data on the user of each device I think the story is different. Imagine what is now mobile, tv, print, outdoor billboards, video game consoles, etc all available through an automated exchange.

    As this inventory consolidates the real challenge will be using all of the data and new targeting abilities to deliver the right message and build your brand depending on the context that user is currently in. Brands that understand this will ensure a quality experience for the user throughout every step of the process so that when it comes to the final purchase decision they stand out.

    Thanks for the feedback (and to everyone else who has commented)

  15. Michael Andrew from Mediasmith, February 19, 2010 at 5:54 p.m.

    Kevin -- I understand and know of marketers who are deep into this (we do much of it) but it has not fully replicated to the entire market yet. The Exchanges are rapidly building impressions and I think by the end of this year we will all be amazed with how much total inventory is in the system.

    On the data front there are many great companies building this out but I sense it is more like search marketing when Google first started up. We are still a few years out before the entire industry settles on how to price/use/evaluate data from the vendors and comes to certain standards, classifications, and best practices.

    For your blog if I knew that you had a high composition of advertising executives consuming the content and could target ads against it that would be worth something. Same for all of the niche content out there that generally gets fairly worthless ads at the moment and far undervalues the actual readers who are engaged with it!

    And I completely agree data is not magic by itself! There are no algorithms that can understand all of the why's implied in the data. That is up to all of us to decipher and act accordingly.

  16. Richard Frankel from Rocket Fuel, February 20, 2010 at 12:17 a.m.

    Michael -- excellent summary of the new emerging world order of data.

    Figuring out how to make use of the data efficiently is certainly the hard part in all this. But even small steps forward would surpass the limited manual efforts that are still typical.

    Plus, we'll be able to attack previously difficult questions -- like how to understand the linkages between DR and brand marketing, long term vs. short term engagement and response.

    I do think that the impression won't become valueless. Just the opposite -- good impressions (pages, audiences, ad environments) will become more valuable as their value is determined based on real correlation to long term business growth for advertisers.

  17. Peter Burgess from Tr-Ac-Net TrueValueMetrics, February 20, 2010 at 1:29 a.m.

    This is a very good thought provoking piece ... but assumes too much, methinks about the importance of an always-on lifestyle and totally connected working mode. The financial sector implosion happened in part when money metrics made no sense relative to the real people and real houses and real incomes. So it goes for everything else in this world where hype too often trumps substance. My guess is that real things are going to have way more importance in the future than they have in the past few years ... with perhaps 4 billion people overly poor and underfed, undereducation and underhealthed ... there are very real things that need to be done ... and we need the creative financial sector and the media world to figure out how to fund and monetize activities that have big value for the future of the global society!
    Peter Burgess
    Community Analytics (CA)

  18. real gdog, February 21, 2010 at 6:49 a.m.

    I agree somewhat.

    No doubt data is important, but so is reach, it always will be.

    The tortuous twists, and contortions in data analysis you imply are simply not as sophisticated as they are made to be, and may never be (excluding search engines at least). The more interesting question is to what end data is used, and with DR only so much can be done, and it doesn't always scale well.

    In terms of immediate selling and acquisition of customers Google is the dr king, and it is yet unclear whether anything other than search intent will every approach it's scale and effectiveness. That Google just bought it's first super bowl ad is an interesting irony, but in doing so joins the super bowl / internet company advertiser ranks with Monster; one of the original online search data companies virtually "made" by the greatest hail mary pass of a super bowl ad of all time.

    Data is important, but reach is just as or more so. I understand media agencies fascination with the supposed new data wave, and do not begrudge them for it, but sometime the more things change the more they stay the same.

  19. James Sandoval from DoublePortrait, February 21, 2010 at 8:27 a.m.

    Wow. I need to re-read this once or twice again to fully digest it, but, for now, very nicely done Andrew.


    I particularly like this: "Looking to find your customers via remarketing? Add yet another tag to the site and give another vendor your valuable customer data".

    That's absolutely right - it floors me that advertisers blindly accept this practice. Something I call "tagging for targeting".

  20. James Sandoval from DoublePortrait, February 21, 2010 at 10 a.m.

    Wow. Very nicely done Michael. I'll re-read a couple of time to digest and maybe come back for another comment.

    For now, I'll say that I really like this "Looking to find your customers via remarketing? Add yet another tag to the site and give another vendor your valuable customer data".

    It floors me that advertisers continue to [blindly] accept this practice of what I call "tagging for targeting".

    A question for advertisers who engage in tagging for targeting practices: How much data is ok to give away for a little [and, arguably, badly measured/attributed] ROI?

    James Sandoval
    Invizua Limited

  21. Scott Pannier from 47 Media, February 22, 2010 at 12:06 p.m.

    Fantastic article and for sure a must-read. I have done some publisher development for some of the user re-targeting companies out there, and impressions truly do not hold much value for them. As an everyday Digsby user, I have to wonder what they are doing with my data, as they have to be selling my data. How else are they paying the bills to keep their lights on?

  22. Jim Ewel, February 22, 2010 at 1:40 p.m.

    Michael - I'd echo the comments of many - fantastic article. I'd also expand upon a point that you hint at. Data will be valuable not only as it goes beyond today's behavioral targeting to reveal intentions, preferences and actions, but also as it allows advertisers to arbitrage anomalies between pricing and value in targeted ads. In other words, not only do we need data to figure out how to deliver the right message at the right time to the right person, but also at the right cost. Today, particularly in buys through ad networks, there is information asymmetry. The networks have more information than the buyer. But this will shift, and networks will be forced to provide more transparency, which will allow smart buyers to use data to reach the right users at the best price.

  23. Ignacio Fanlo from AdBrite, February 23, 2010 at 11:58 a.m.

    Almost every time an article is this long, there is fluff. Not this time. You nailed it.

    I'd make two comments. First, AdBrite is an online exchanges and while we use that nomenclature, I'd say that exchange is just good short hand. Ad inventory is perishable and as such these "exchanges" are more like eBay meets Expedia/SABRE; an auction on top of a GDS.
    Second, unlike media, where the impression can only be sold once, data can be sold an infinite number of times by the owner; similar to an airplane seat versus the data on who bought the plane seat. Therefore, the value of data will eventually fall below the value of the media at equilibrium. It's the smartest USE of the data that should trump all. In the end game, content will naturally reap the greatest benefits, finally being king again.

  24. Lee Freund from TubeMogul, February 24, 2010 at 3:10 p.m.

    With all the digital data talk, it's terrific to hear someone mention creative.

Next story loading loading..