Counterpoint: Data and Media Work Well Together – Separately

Networking“Networking” is written by members of the online advertising network community.

Today’s column is written by Mark Zagorski is CRO, eXelate

I applaud John Garber for taking this subject on in his very well written and interesting piece (read it) yesterday.  Although I think many of the concepts are right on, including the nascent nature of data valuation models and the added value of context+audience in determining campaign success, I have to take issue with a few of the points, which I feel undermine the general premise of maximizing the value of audience data for publishers, and which ultimately may confuse media buyers.

First, the idea that “the risk is all on the buyer in a straight data buy” is true only in models in which data is purchased on a “per user” basis and it is then up to the buyer to find these users either via their own direct media sources or media aggregators. The pay-per-user model may work for buyers in some cases, but for the increasingly real time, DSP-console driven nature of media buying it doesn’t make sense. That is why eXelate (and other data companies) also built data models on a “pay per use basis”, in which data costs are only incurred when a user is “found” and successfully targeted. In these cases, there is no buyer risk, and the publisher (data provider) shares in the upside on every impression when its data is successfully used. In this manner, publishers are not the guinea pigs in a “data buyer arbitrage” game, but active participants in the sale whose value rises as based on the ultimate worth and incremental use of their data.

Secondly, the idea that media and data need to be smartly coupled to deliver value is right on the money, but the underlying assumption which is being made (I believe) is that the player managing both assets needs to be the same, and that if the data provider is not also contributing media to the “data engine” then there is a value loss. This to me is not so clear. Buying data and media together in a model that rewards performance and reach is not mutually exclusive with publishers contributing media and data to the same partner.

A few months ago, Ad posed a question – “Should data and media be sold together by the same company?” As expected, all of the key players chimed in, each with their own take on the question and each with a self serving (myself included) view on what the best direction should be.

At the time my take was that I believed that there is always some level of conflict when you aren’t a pure player in any sector, because at the end of the day you either (a) end up being competitive with your own customers or (b) fail to fully serve their interests. The key reason that data and media access were split in the first place was because of these inherent conflicts, and a main reason why independent data companies are thriving.

My position several weeks later (that’s 5 years in internet time 😉 ) remains the same. With the proliferation of media platforms, and the ability for smart data companies to integrate into these platforms, there is no need for data companies, and the publishers who work with data companies, to get into the media business together. Contextual filters on DSPs and media exchanges enable the type of targeting that can enhance the performance of data thus driving higher ultimate value for the data provider. Plus, when data and media are combined in these separate platforms, data source names are anonymous and data pools are aggregated in order to mitigate channel conflict. So, there is none of the uncomfortable concerns that the guy sitting next to you in the waiting room of the media agency is selling your own audience against you (and knows exactly who you are).

Additionally, a fast way for publishers to lose data value is for a current data buyer to walk away from a data platform or provider because they are now in the media business. As one of our partners quipped recently – “Let me know if you guys get into the media business, because if you do, this is our last lunch, even if you are buying!”

Because of business model variations and the potential for channel conflict, I think it is too early to make a call for the combination of data plus media from the same sources in the same marketplace. In our rapidly growing online advertising ecosystem, where we see company models change at will and whole new market segments spring up overnight, my take is that publishers should carefully consider if this direction is a short term branch in the evolutionary chain or a long term shift in market direction, before they take the leap.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Paul Silver

    I think one of the biggest challenges (from the buy side here) is how to demonstrate the incremental value delineating data from media provides – in my experience so far(and others I speak to), this is not demonstrated in a drop in CPA or increase in ROI. Considering most budget through DSPs is from DR advertisers, this seems to be a pretty big issue. Would love to hear from anyone on this note.

  2. You say “there no need for data companies, and the publishers who work with data companies, to get into the media business together” – that is fine. However you fail to look at this from another perspective. There *is* a need for publishers to get in the data business. This speak directly to your premise of “maximizing the value of audience data for publishers” and “confused media buyers”

    Kawaja spoke about publisher “branded data” and this is but one way to maximize value. The main issue is if publishers continue to take their data inventory and hand it to third-parties for monetization the same way they handed networks their ad inventory, the same commoditization, devaluation and loss of revenue will ultimately occur.

    Publishers should be wrapping the inventory with data themselves and the price should reflect the quality of both and the needs of the advertisers. The data will be better for it, the media will be better for it and most of all the advertisers will save a ton of time and energy trying to piece together successful campaigns in an ever more fragmented and difficult market to optimize in.

  3. Mark Zagorski of eXelate wrote a very good counterpoint to my comments around pairing publisher data with publisher media in order to maximize revenue opportunity for a publisher. After reading it, I was struck with two thoughts: 1) Mark and I pretty much are in violent agreement on the majority of the conversation’s substance. The buyer needs to come first in an early stage market in order to drive long term demand and that can’t be done if the buyer is bridled with the responsibility of monetizing data on its own; and 2) as with most such debates, there is a ton more nuance to this topic than I had originally addressed. So I thought I might, hopefully, provide some clarity.

    In keeping with thought number one, I absolutely agree that if you’re going to buy data separate from the media, it makes the most sense to buy it in a pay per use model. Until the demand side grows, it’s the best way to go. My message in the original article was aimed solely at setting publisher expectations when valuing a publisher’s data and engaging with buyers to sell it:

    Dear Mr./Mrs. Publisher,

    You’re going to have to sell data on use. You’re going to have to incur the risk. And, unfortunately for you, that’s going to create some hurdles. For one, you’re not going to be able to reliably predict how much you’ll sell. You’ll also have little insight into why the buyer bought what they bought, whether or not it helped drive a successful campaign or what you could do to make your offering better and more valuable.

    The Data Market

    As someone who both buys and sells data, when selling on the per use model, my next quarter revenue predictions are, well, let’s just say “muddy” at best.

    All that being the case, it’s a nice segue into the second topic around the value (or not) of selling data and media combined. Just to be clear, I’m not proposing that a publisher do one and only one method. I’m not even proposing a publisher sell data to only one reseller. What I am proposing is that the value of data is largely driven by the opportunity to place a targeted advertisement in front of a potential consumer. Without that opportunity, value deteriorates.

    Think for a moment of how many web sites you regularly visit. I’m not talking about the “I wonder what day Desperate Housewives starts back up next month” surfing… I’m talking about sites you visit every day for regular content. For me, I would say it’s five, max (and several other N > 1 studies on the subject of ‘how many sites do you visit in a month’ confirm that it’s far less than you might expect, typically < 10). The majority of opportunities to present me with an advertising message exist on just one of those five to ten sites.

    So when you think about data and media as separate entities, the amount of opportunity to use data (even when you don't pay unless you use it) is really driven by the crossover between the data sellers and the media sellers. From my experience of buying and selling data, I'm happy with 15% – 20% crossover when they are separate. I get closer to 90%+ when I also buy the media where the data was generated too. Again, I'm not suggesting you do one or the other, but I am suggesting that unless your data segment is massive, a large portion of the data insight goes unused. Oh, and if one of those five sites you surf every day is Facebook, the data and media buying plan gets even more complicated. Nuance is complicated.

    With regard to Mark's position on whether or not it's best to sell just data or just media or both, I agree. (Wow, these debates aren't nearly as fun when we keep agreeing with each other!) It has really yet to be determined. Full disclosure: Lotame has both data and media arms. I'm of the opinion that the two go hand in hand, not only for the reasons I listed above, but also because of the added value that running campaigns on the media brings to data segment creation: When media and data are coupled together, one can see what data did or did not work for an advertiser. Because one gets to see what data drove creative interaction, that learning can be applied to package better and sell better data segments. Does that mean it's better to have both? It’s still too early to say, but I remain as convinced as I was from the thesis presented in my first article: if you're a publisher and you want to maximize what you can make from your data, I still think pairing the two elements will bring in the most dollars.

  4. Good points, all. And exactly the kind of smart discussion that helps the industry better understand just what the heck is going on. Plus, any comment that includes a “Dear John” letter goes beyond debate . . . it is pure literary enjoyment. Great work, Mr. Garber!

    I won’t adress all of the counterpoints individually, but agree on one general premise which I think underlies both the media bundling concept and the direct sale of data — that of data provider transparency.

    Although not for every publisher, there is value in a direct, transparent sale of data. That is why eXelate built our Private Marketplace, so that pubs could sell premium data directly to advertisers, while still having the option to anonymously bundle targeting information in a blind marketplace as well.

    Transparency always drives higher value, so not the end game in every case, but definitely an option.

    As we work through all of the best practices in this truly nascent market segment, opinions will change (this is my hedge!) . . . but with the rapid dissemination of information the one deciding factor that will form the debate will be which model creates the best value while protecting the core business. May the best model win!

  5. I recently wrote a piece on this site on the power of placement and the likelihood that data prices are hugely inflated based on the fact that advertisers are charged for “reaching” a user regardless of whether that user can actually view the ad. No response from the data players – and no mention of placement in the debate above. Are you hoping if you ignore this inconvenient truth it will go away? Or do you believe it’s a non-issue, and if so, why?

    • Kirby – Please clarify your point. Why do you believe advertisers are paying for the data regardless of whether the user can see the ad?

  6. Mark Zagorski

    Alan summed it up well.

    1. In many data models you only pay when the correct target audience views the ad.

    2. Context and placement are absolutely important and work hand and hand with data.

    Think of data as the no risk, high power optimization accelerant. It’s like a squirt of lighter fluid on a hot grill!

  7. Alan and Mark,

    Thanks for responding! I should have clarified by including this:

    Basically, when a marketer purchases media, regardless of whether data is included, there is a percentage of that media buy that is never viewed by a user (below the fold and not scrolled into view, etc.). And the standard amount of wasted impressions in a campaign costs the marketer X dollars.

    So, since when a marketer purchases data and uses it to target advertising, it increases the CPM cost of media by some factor (call it 5), the wasted X dollars becomes a wasted 5X dollars. The premise is that overlaying and paying for data based on reaching an audience that may or may not be able to view the ad causes exponential wasted budget. It’s lighter fluid on a hot grill alright!

    Make sense? Or am I missing something?