One Question: How Does Programmatic Buying Become Attractive To The Brand Marketer?

One QuestionOften, a question doesn’t have an easy answer in the digital advertising business.  This is a new column devoted to an answer to a single question – and providing a bit of space for it.

Today’s participant is TRAFFIQ Chief Product Officer Eric Picard who recently answered the following question during a conversation with… How does programmatic buying become attractive to the brand marketer?

EP: I love this question. I personally believe that all of the interesting things that are happening in the online display advertising space relate to audience-based buys designed to reach specific audiences -which are basically premium brand buys.

Display advertising, while it can be used for performance, it can only really be used well for performance when the inventory is incredibly inexpensive. It’s just a basic, simple fact that performance ad buys either have to drive very high conversion rates, or they have to be very cheap and you can see it in every media to some degree – and especially in online display.

Brand advertisers are willing to spend significant amounts of money to efficiently and effectively reach potential customers that fit the profile that they’ve developed around their target audience. The reality is we talk a lot in this industry about accountability, the ability to control the way that ads are displayed as if it’s new and interesting.

The VP of Marketing

One of the great things about my experience working at Microsoft is that I received access to everybody. I was in many, many meetings with some of the top brands in the world, I’d be able to just ask them questions about their advertising, the way they made decisions, why they spent more on traditional than display proportionate to the time spent by audiences. I recall one fascinating conversation – I won’t name the name of the company – but the brand marketing team are in the room and some online display media sales people came in. The sales team tells the usual story about accountability and the ability to control the way the brand’s budget is being spent at such granularity. The marketing VP became angry and frustrated.

He said something like, “I don’t want to hear any more from you online media sales guys about how accountable you are. Do you think we’re idiots? We spend billions of dollars a year on ads. And that the whole thing about half my ad spend is wasted and blah, blah, blah –well, I don’t want to hear anymore about that.”

And then he said, “First of all, I know exactly what my ROI is on every dollar I spend, down to multiple decimal points of percentage. There’s no question about it. We spend a huge amount on research, and we spend a huge amount on validating our ROI. We’re very, very confident in all of that stuff. So, please don’t talk to me about how accountable you are. I actually have found that online media is not accountable at all.”

He added, “The biggest problem that we have with online media is inefficiency. Online media is between 10 and 15 times less efficient to buy and to manage than traditional media. That leaves me with this problem, how do I justify spending any money on online display when it’s 10 to 15 times less efficient to buy and manage. That means that I need to find media that’s either 10 to 15 times less expensive, or I need to find ways to get 10 to 15 times the ROI. But neither of these is true today.”

A lot of people don’t remember that when we started building ad platforms at the beginning of this industry, it was a very nascent industry. A lot of the business rules we put into ad platforms were designed around the needs of a nascent, small -very small -budget, highly manual industry. But then, we never went back and rebuilt those platforms.

The amount of human effort required to manage an ad campaign at a large agency for a large account – take a Proctor & Gamble, or a Johnson & Johnson, or any of the companies that spend huge amounts of money on ads across all media – they don’t want to be worrying about per placement optimization. That’s way beyond anything that they ever need to do in traditional media. What they want is to be able to specify an audience, to figure out what properties to buy it on and then very efficiently spend lots of money. Online is just so inefficient, it’s so much more labor intensive for them to manage that it doesn’t make sense to them.

I think there’s an opportunity for this industry to shift the large brands dollars online, for the first time, because it’s about efficiency more than it is about anything else. I need to be able to efficiently reach my target audience, at scale, and buy at scale as a large advertiser, and I don’t mean I want to reach a thousand people on this site, and a thousand on that site. I want the kind of scale that you can get from television or magazines, and do that without a lot of human intervention. If I can do that as a buyer, then we’re going to see this market shift very fast.

The problem we keep running into is there’s a lot of legacy mindset. People are still very focused on granular knobs and dials to manage and manipulate campaigns. It’s fine if the systems are smart enough to do that automatically, but if people have to get involved and approve, or look at, and manage, and go through all of the rigmarole that they go through on a typical display campaign, there’s just no way.It will be the death of this industry.

We won’t ever get out of the hole that we’ve been in for the last 15 years. Which isn’t to say there isn’t real money spent here, but it isn’t the kind of money that we could be getting access to, and it’s very disappointing.

The Arrival Of Programmatic Buying

Programmatic buying systems are starting to be put into place today, but I think the bigger problem is the kind of inventory that’s available for programmatic buying today. There’s not enough of it. This is a really tricky point. I’ve had a lot of debate back and forth on this point so I’ll just spend a minute on it. There’s this belief in this industry that there is infinite amounts of display inventory in this space and the reality is there isn’t infinite inventory, but there is a lot of it.

There is a lot of bad inventory, or weak inventory, or “not great” inventory which is exposed into these real time bidding environments and where the inventory is either sold completely blind -with whatever targeting attributes a demand side platform, for example, can bring to the table – or perhaps little bit of targeting coming from the publisher.

Part of the problem we’ve got is that publishers are so sensitive about the kind of change that could happen in this space that they’re not even willing, in many cases, to experiment with putting the inventory out there in a way that is useful to the buying community. So, until we can get to that point, I think we’re going to continue to have debates and a lot of issues.

And, I think the problem that we’ve got is the power that the individual sales forces have, the amount of say they have in strategic direction of the publishers, and the general concern that those sales forces have about their relevance as a sales person or a sales team in a role going forward and in a world where people can buy without negotiating.

But in reality the role of sales will be in many ways more important – it will be evangelical. It will be about convincing the buyer to put their site into the business rules of the buy such that they’re a must-buy. So even if the inventory isn’t negotiated or taken by sales – and even if the RFP isn’t answered by a sales person, they have a hugely important role to play going forward. But the role will become more strategic, evangelical and will drop all the tactical work like RFP response.

Removing The Risk

Also, the next phase of that is how do publishers make their inventory available programmatically without taking risk on revenue. I actually think what we’ll find – and I’ve done some modeling on this – we’ll find that a lot of inventory will be worth zero, which it almost is today anyway.

But, the most important inventory, the inventory that’s desirable, that’s attributable to desirable audiences, will have a lot of competition. The competition on those audiences will drive the price up. And the brand advertisers won’t mind paying a higher overall price as long as the buys become much more efficient to execute and manage. They don’t want the ‘zero value’ inventory anyway.

Publisher systems have been slow to add in some of the really important things that have to happen, like real time floor pricing on a per impression basis.

Instead of doing inventory prediction so they can sell far in advance, they should be doing inventory prediction that looks at what the likely range of prices is on available inventory over the course of the next 10 minutes, so that they can actually start predicting what the floor price on those impressions should be.

I think we’re in a world where there’s such asymmetry between the goals of the campaign, the value of the inventory, and the willingness of advertisers to pay, that second price auctions are going to cause a lot of problems, especially as we’re growing.

And, where you have one advertiser that’s willing to spend a $50 CPM to reach that impression, or the person behind that impression, and you’ve got another advertiser who’s willing to spend $20 and another advertiser who’s willing to spend $3 and one who’s willing to spend 50 cents, and the gap between the highest priced bid and the second price is significant enough that we’re going to have to figure a few things out.

So, if I were building a publisher-side system right now, I’d be thinking really carefully about that problem because it’s a big problem. And, it’s one that, once you get that solved, then a lot of your concerns go away because you basically are confident that you’re going to be able to get the price that you need in order to move more and more of your inventory into those models.

Follow Eric Picard (@ericpicard), TRAFFIQ (@TRAFFIQ) and (@adexchanger) on Twitter.

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. Great piece Eric (I like the new format John).

    I think a lot of the concerns you are expressing about the quality of the inventory on the exchanges are rapidly being addressed as the average exchange cpms increase as a result of efficiency gains of RTB. As we have seen the prices rise in the last year from twenty-five cents to almost two dollars now, the quantity and quality of the inventory has increased exponentially. The hockey stick growth of RTB can be understood simply as millions of publishers swapping out their ad tags to realize the increased yield of exchanges vs selling to networks. We at Triggit expect the average cpm to cross $3.00 next year. When that happens it will be very hard for a large proportion of publishers not to sell most if not their entire inventory to automated channels. The publisher sales forces will certainly always fight a rearguard action to attempt to forestall their jobs from being automated, but as prices on the exchanges increase it will become virtually impossible to justify all but the top sales organizations selling anything but sponsorships and custom campaigns.

    • Zach – I’m curious how the addition of hundreds of millions of units of inventory will drive up prices. The main thing I’ve seen drive up prices in most markets is scarcity. Search is a great example here – there’s only a limited amount of impressions each day on the top performing kws. What’s the analogy in RTB?

      To Eric’s point the efficiency gains you speak of are on the buy side. Is it not possible that publishers will discover that most of their inventory is indeed worthless in an open marketplace – especially one that (at the moment) relies on third party data matching to effectuate the value?

      Really interested in your POV on these questions.

      Eric – really great piece…thankfully John has increased the word count max to get this all in. Couldn’t agree more that a legacy mindset is holding things back. I’d take it one step further that it truly threatens the viability of the channel.

  2. Eric,

    Very well written, insightful and timely article. Thanks for sharing your interesting experiences with brand marketing executives.

    Totally agree that the predominant ad servers of the day have people trapped in a time warp from the late 90s. This is definitely a significant contributor to the inefficiencies, along with the creative scalability problem – which both need to be addressed.

    If we define the problem of campaign executions in the following way one aspect of the solution emerges:

    – WHO to reach: Media planning, buying and optimization is about articulating the Audiences desired (demographics and intent) and having big enough aggregated inventory sources to find them.

    – HOW MUCH: The budget

    – WHAT: The types of creative experiences, product listings, offers, rich media assets and messages to show to the audiences in a real time application delivery mode, and not as static files

    – HOW’S IT GOING: As transparently as possible, show the performance, pacing and reach of the campaign.

    – PERFORM: by applying optimization methods, whatever they may be to reach the metrics of the campaign.

    So, we need an Auto-optimizing, Automatic media buying platform with a Personalized Data Driven Creative Ad Server along with Near real-time Analytics.

    Now, for the shameless plug:

    With the exception of the “automatic media buying” piece, the rest of the infrastructure is what we have in the Tumri Dynamic Creative Ad serving platform. And our deployments are all in partnership with DSPs, agency trading desks, as well as direct clients executing to this vision.

    Thanks again for an excellent article.

    — Pradeep Javangula
    Founder & CTO, Tumri Inc.

  3. Eric – thanks for a very interesting piece!

    Zach – thanks for being optimistic about the future of what we’re all up to! 🙂

    I might be missing something completely obvious by I disagree that “more efficiency” in terms of managing distribution overheads is what will move the needle in getting online media branding to be on par with online media usage.

    My perspective on this issue is very different than most guys here on AdExchanger. I cannot make a solid argument for the difference between DSP, DBO or a trading desk, but I was the only ad tech guy doing a non-paid-for speech at this year’s Cannes Lions 🙂 My background is technology+creative agency and (though we’re not talking about it that much) my current company, Burt, actually has most of the world’s top creative shops signed up. We don’t have an explicit focus on “branding” I would guess that 75% of our business is brand ads.

    Moving on…

    Let’s just say that the reports we all read are right. 30% of online budgets are spent on distribution overheads (planning, buying, traffic etc). And let’s say some magic product comes along that eliminates this cost *completely* AND you’re able to reinvest all that money with the same ROI (which is rarely the case, and even more so with impression level granularity buying) this would be your maximum lift.

    And sure, you can always reduce waste in your media plan thru audience buying, or doing quantitive optimization of the creative execution. But these things hit a local maxima pretty soon because online ads – as they’re made, measure and placed today – simply don’t have the -*impact*. We distribute crappy ads that appear in placements where most viewers never care to look. Distribution optimization won’t save us from that fact that audience just isn’t paying attention.

    Eric – precisely as you say, any half decent brand has had their attribution models nailed since forever. And when the numbers come in they see that online just isn’t delivering the shift in perception they see.

    What we need in order for the new distribution paradigm to flourish is better fundamentals – better ads, better placements, and meaningful metrics that explains how we can improve even more. Optimizing the flawed distribution model is important, it will not be enough to shift brand budgets online in any meaningful way.

    Enough snake oil 🙂 Back to coding!

  4. Gustav, at the risk of reading too much into Eric’s piece and interpreting what he is saying – here is my 2 cents:

    a) Removing distribution challenges, the overhead and the teams that mushroom around it would enable better focus on actual marketing needs: i.e., better strategy, better planning and overall more impactful business thinking.

    b) Crappy ads and suboptimal placements should just automatically be eliminated given rigorous KPI requirements. It should be the case in an evolutionary model. The fact that anybody needs to pay attention to these details is in and of itself a problem with the ecosystem.

    c) Let’s not dismiss the fact that there are only so many “great” ads that can be produced by creative executioners. Spray and pray models definitely don’t work. What has reliably worked for web front-ends are personalization methods a la recommendations, data driven presentations etc. So, the model in display should definitely go down that path.

    d) Attribution model is far from nailed. Especially cross channel attribution. The underlying population and the context of ad serving is of an extremely high dimensionality, and what impacts what is a research problem that needs to be solved technically first for scale, along with business acumen.

    I think the order is something like:

    – Fix distribution models, i.e., the pipes for creative and media.
    – Adopt rigorous metrics and solve the attribution modeling problem at scale
    – Understand and address ever increasing and increasingly complex distribution modalities and touch points for consumers: mobile, video, social etc. etc.

    Then, when we can say, that we have the infrastructure laid out for efficiency – we can ask brand marketers why they think it is wise to have this lopsided marketing mix between offline and online.

    But then, I am pontificating as much as the next guy. What do I know?:-)

    Happy thanksgiving to all!

    –pradeep javangula

  5. Jonathan and Gustav, all media has an inherent value at the impression level that is independent of its relative scarcity. At some price every impression has a buyer regardless of how many other impressions like it there are on the market. . To the extent that media can be used to further quantifiable business results then that media value can be discovered and media can be priced differentially. Eric is certainly right that a large amount of media currently on the market is not worth much, but rationalization of pricing in the market with RTB should help publishers recognize this and get them to stop selling the worst junk. Our goal in this ecosystem ought not to be to drive up false scarcity but instead to improve the value chain to increase that inherent value and remove unnecessary inefficiency. Basically, I would rather build better things then spend my time figuring out how to extract bigger tolls. My argument about driving up prices is that as we build better technology to improve a stupidly inefficient system and we built better tools for determining media value we will increase that measureable inherent value and drive up prices. To the extent this was predicted over two years ago and the media prices on RTB have kept climbing I am comfortable that this is the correct way to think about it versus the sky is falling counterargument about lack of scarcity.

    Gustav, I completely agree that we need bigger and better ad units to speak to consumers (I wrote about it briefly in a post on Adexchanger a couple months ago here)

  6. Thanks for the lively comments. I’ll say a few things in answer (summarily) to what was posted:

    There is a vast quantity (growing daily) of crappy worthless inventory. This worthless inventory that most publishers are monetizing well below $.25 CPM (I’m hearing from some remnant buyers that their *average* CPM for inventory is $.07). At these low (almost free) prices, direct marketing models can work – but the vast amount of crappy creative flooding the market isn’t helping our case.

    Taking this $0.05 CPM inventory and adding targeting can increase the CPM significantly (where targeting can be applied) – like into the $2-3 range on average.

    But this is still super cheap.

    The goal I’d set is to get out of the remnant pool model – only sell valuable opportunities to reach an audience and don’t show ads otherwise. Limit remnant to 10% of buys and some PSAs. But also decrease ad clutter to increase attention to the ads.

    Let’s talk about TV for a minute…

    Network Television CPMs (Primetime) average (depending on what source you use) about $12-15. Keeping in mind this is fully diluted – so there are many buys in the $15-20 range. And a lot of unsold too.

    Average Television CPMs across broadcast and cable are in the (again depending on source) $7-9 range. That’s for Full Motion Video with sound even (talkies anyone?).

    Note that most TV Stations and Cable Networks limit the amount of remnant they sell to ~10% (this goes up during bad economy)to keep a lid on bad / irrelevant creative.

    The thing about TV is that planning it is super easy, and buying it is super easy. Planning TV is easy because every single show, every single pod and pod location is known to everyone well in advance. Planners can rip and replace shows they want to buy and use GRPs or TRPs to normalize against similar shows. Buying is a simple negotiation.

    In online display the publisher sales force keeps all the information about the buy close to the vest. The RFP process is a nightmare for both sides of the table, but publishers are fearful to give up the control they believe they have. In reality cost of sales is through the roof online, and cost of managing buys both in planning and after a buy goes live are out of control.

    This is NOT a zero sum game. Advertisers see the higher costs of managing, and simply don’t put any significant buys online. So it’s not JUST the cost of current media plus whatever percentage we can add to today’s base CPMs. It’s the amount they’re willing to pay to reach an audience in our media. Once the gates open, the flood of quality buys at relatively high CPMs will follow. But for now, we’re locked out due to the vast differences in methodology and inefficiency of our processes.

    If we really want to have things take off in this industry – we’ll increase the opportunity to view audience that is valuable through programmatic buying, and we’ll decrease the amount of crap remnant inventory we sell below a certain floor price.

    The amount of yield increase for publishers that we could get by selling more higher priced inventory at higher levels of efficiency more than outweigh the loss of remnant inventory sales.

    One last thing… The amount of valuable inventory in our industry is limited by the number of valuable human beings that match advertiser campaign goals that are visiting pages. The number of people in the world who are good valuable targets is only increasing or decreasing according to the population and incomes of the audiences. RTB of whatever flavor allows advertisers to bypass the crap. Which they don’t want anyway. So all those worthless impressions that we basically can’t really sell today anyway just don’t get sold – which should be a good thing.

    Eric Picard

  7. Hey guys, thanks for straightening some things out for me! I guess my gut feeling is that ad tech is often guilty of trying to boil the ocean 🙂

  8. Good article but I have to say your comment about Performance advertisers is actually wrong. All we do is run hard target DR advertisers. Click to conversion, no latent clicks, direct path referrals, where real de-duped leads that turn into customers need to show up in the database. (I know Zach is having convulsions right now that we would still consider marketing like this) The fact is we get things to work. We get things to work by blasting a billion impressions at pennies on the CPM which is what you are alluding to. But we also buy homepages and buy premium placments and have paid from $10-$100 CPM and hit hard DR targets. The fact is people are extremely lazy when it comes to creative and they don’t know how to optimize. If you do that well, you can get high CPM placements to work for hard target DR advertisers you just have to know what you’re doing. What I would love to see is some of those big huge premium placements in a biddable and schedulable environment. I don’t care if you set the floor price at $30 CPM, if I can schedule it and optimize it and adserve it and track it, I will work the creative and landing page process and get it to work.