Konrad Feldman is CEO of Quantcast, an online advertising solutions company.
AdExchanger.com: Overall, what industry-wide trends are you seeing across your product lines in 2009?
KF: From our vantage point we are seeing tremendous energy by brand advertisers and their publishing partners to unlock the value of real-time audiences. At the end of the day, brand marketers want to reach their customers and prospects at scale – with limited waste. If publishers can deliver against that objective, they can drive their own revenue yield, and expand their advertising base. It’s potentially a win-win for the marketplace, but it requires some fundamental shifts in how we operate.
Digital media distribution creates opportunities and challenges: consumption is increasingly fragmented and impression delivery is allocated dynamically, but there is no market-wide solution to deliver advertising against audiences in real time. The result has been a brand-based online ad model grounded in total impression – not audience – delivery. It’s hard to know what actual audience is bought/delivered when the yardstick for evaluation (historical property-based audiences) is completely disconnected from the delivery model (real-time impressions). Just because you “buy” a million impressions on a site with a historically high male skew doesn’t mean the impressions you have delivered in real-time are actually the men.
We’re excited by the rate of adoption of the products we’ve launched in 2009. Quantcast Marketer helps agencies and advertisers understand the distinctive characteristics of their best audiences and Quantcast Media Program allows buyers and sellers to execute addressable media buys using these audience definitions as a consistent yardstick for impression selection and delivery. These products just launched in Q2 and we’re already seeing significant active participation from leading players on both the buy and sell side of our industry.
In AdAge you said that your new Media Program “allows marketers to define their customers … rather than a media entity making that package or translation for them.” I believe you’re referring to behavioral ad networks and publishers here. Can you elaborate on what is getting lost in the translation and how Quantcast improves it?
Think about it this way – to a certain extent, media planning, buying, and delivery has been like a game of “telephone”. Because the process is disconnected, there is significant degradation of value from start to finish. Quantcast connects the process, and maintains as much value as possible for buyer and seller.
Marketers, for decades, have had large volumes of both proprietary and syndicated intelligence about their consumers – sales, customer touch point, loyalty program data, etc. In a digital world that data set increases, through customer interactions with advertising, paid search links and brand web site activity. Both offline and online, marketers and their agencies use this information to build deep insights and segmentation models about their desired target audiences. But ultimately, when placing advertising, they’ve had to evaluate and buy media based on generalized audience attributes – because that’s the marketplace currency. Sure, there have been ways to target based on behavior, or action – think coupons you receive in the supermarket checkout line based on something you just purchased, or delivering an incentive based banner ad to someone who abandoned an online shopping cart. Those are tactical solutions. From a strategic perspective – driving scale, and more importantly new prospects – brand advertisers have used broad content or demographic descriptors like age and gender to evaluate and buy audiences.
Online, the problem is more complex. While there are a host of offerings that enable more specific targeting (millions of niche sites, BT as you suggest, or registration-based targeting) the problem becomes one of both segment relevancy to the advertiser, and of consistency. What do we mean by segment relevancy? Well, there are a vast number of segments brand advertisers might be interested in at any one time (driven by whatever marketing objective is in play, for example share/shift goals drive a different targeting strategy than driving usage/frequency by loyal customers). But it’s not practical for the sell side to create and manage a vast number of audience segments. So the marketplace has created – in a fairly ad hoc way – a variety of ways to target, none of which deliver the exact needs of an advertiser at a particular time. From a consistency perspective – when you have thousands of ad sellers each delivering mutually exclusive targeting offerings from a potpourri of arbitrary, content genre definitions, self-reported demographic information and third party licensed data, you lose apples-to-apples comparability from one seller to another. Who knows if Site A’s “automotive” segment is of equal quality and precision to Site B’s, or whether Site C’s auto racing content delivers the same type of enthusiasts as Site D’s? Even more importantly, the buyer is left to translate this hodgepodge of audiences back into a form that can be compared against the desired audience segment they have spent so much time and money developing.
Our belief is that the model needs to be flipped. Advertisers need to be able to use their own data to build customized audience definitions representing who they want to reach, and that “consumer profile” needs to be actionable across the web at scale. This approach, solves two critical challenges that brand advertisers continue to confront as they attempt to shift more of their dollars to digital media: 1) it allows marketer-driven consumer segments to be uniquely defined, and 2) it allows them to be activated at scale, consistently.
What’s your view on ad exchanges?
Ad exchanges create value by generating additional liquidity for display advertising inventory and reducing the transaction costs associated with buying and selling this inventory. Ad exchanges also bring technical optimization capabilities to market participants who previously were unable to access them in-house.
At Quantcast, we see ad exchanges as an important opportunity for Quantcast-enabled marketers to connect with their custom audience segments. By allowing marketers to build actionable audience segments and then leverage those audience segments across the ad exchanges, Quantcast brings consistency and accountability to media buying and liberates advertisers from outdated, index-based, site-centric buying mechanisms.
It would appear that the opportunity to buy “lookalike” consumers – those being consumers that successfully fulfilled a specific advertiser action – will drive huge interest. But how does it work? For example, as target consumers are effectively identified by Quantcast on the exchange, the buyer will be able to purchase impressions only for that type of look-alike consumer. How am I doing? Does the IO go through you, the exchange?
Let’s take a step back. First and foremost, a “lookalike” audience is not simply a group of people that have successfully completed an advertiser action. The premise of our lookalike model is that insights from small (or large) groups of people who take action through marketer-controlled content allows that advertiser to build an aggregate understanding of their target audience, so they can extend reach against a broader, relevant audience. For years, marketers have been able to “retarget” groups of people who take an action on their site – say the 20,000 people who download a coupon. This simply builds frequency against an already engaged group. What they haven’t been able to do is leverage the aggregate attributes – both what those people are, and are not – to find much larger groups of similar people across the web.
You mentioned the “exchanges” in your question, in terms of how lookalike models are activated across the web. It’s important to note that exchanges are only one of many inventory sources advertisers can leverage to deliver Quantcast-enabled audience segments. Our service is built to accommodate whatever inventory sourcing model an advertiser and/or publisher chooses to deploy. The opportunity for advertisers is not to simply source inventory from exchanges, but rather to leverage a lookalike audience across the web. Any publisher, network, or exchange that is participating in our service will be able to deliver an advertiser’s lookalike audience.
In terms of how the process works – it’s quite simple. A marketer “tags” their brand site/content (particular events, product info, even ad campaigns and search activity). They use the insights from one, or any combination of these events, to power the “lookalike” model, and an aggregate audience profile is developed. Quantcast then scores the Web – we help the advertiser understand how many people with a similar set of attributes exist, and the volume of their composition on particular sites. We then enable publishers to sell groups of impressions that have high conformance to the lookalike profile to the advertiser. The advertiser can purchase their audience directly from Quantcast, or they can choose to work directly with publishers if they have existing relationships. Ultimately, we will work with ad buyers and sellers however they want – our goal is to provide solutions.
Will Quantcast Media Program be integrated into all the exchanges and when? Is this a time-consuming process? Assume you are mapping Quantcast cookies to the exchange’s cookies here. Given your scale, this could be a huge task.
Quantcast’s scale is actually a benefit in enabling our customers to act on their audience segments. Because of our broad distribution we can offer a myriad of options for a marketer to gain access to their lookalike audience. Quantcast-enabled marketers can leverage their custom audience segments across the ad exchanges today; Quantcast works with RightMedia, the Google Exchange and ContextWeb. As the exchanges and our entire ecosystem evolve, Quantcast will continue to create integration points for the marketplace with the goal of increasing liquidity for Quantcast Marketers and Publishers.
A basic behavioral/demo question for you: In that user data is anonymized, how does Quantcast, or any behavioral firm for that matter, truly know that a user is a woman, for example?
The answer to the question is that no targeting solution “knows for sure” that a specific person has a particular attribute. Our approach is based not on targeting individuals, but rather, helping ad sellers and buyers transact groups of people that have above average propensity to align to a certain set of characteristics. There will always be some people in that group – for a whole host of reasons – who don’t conform perfectly to the target, or at all. Our goal is to minimize waste as much as possible and enable delivery of enriched audiences that perform for our clients
The reason we’re confident about our approach is because our programs have very broad participation. Over 10 million web destinations participate in our program, and collectively those organizations engage every US Internet user many hundreds of times each month. This provides the visibility necessary to construct accurate fine-grained models of virtually any audience.
If you were running a digital media buying agency, what changes or strategies would you put in place to prepare for the future?
The first thing we should recognize is that future of most media is digital – the lessons we learn online today, will apply to television tomorrow. Agencies, both digital and parent, must transition to a future built on insights, attribution and performance. I don’t mean performance in terms of just clicks, but rather true performance for the marketer, whatever their goals might be. There is a vast amount of data that is now accessible by agencies – but to make it valuable it must be made easy to use, and actionable.
Digital agency execs understand this – and many are focused on retooling their organizations with robust data analytics and modeling capabilities.
I wouldn’t suppose for a moment that I’d be any good at running a digital buying agency, but I do think I’d focus my organization on leveraging technology holistically to support the transactional elements of my business and focus on doing what agencies do best – developing ways to differentiate their offering through unique application of data and insights and creative solutions for my clients.
How will real-time bidding (RTB) and demand-side optimization affect Quantcast?
Real-time bidding will offer Quantcast’s clients new opportunities to apply insights and select the right impressions in real-time for their campaign.
In AdAge it was noted that Quantcast isn’t ubiquitous. On the other hand, you say in your marketing literature that you cover all 220 million U.S. uniques. Where’s the drop-off? Perhaps the point is that you’re not seeing what users are doing all the time, so it’s hard for you to create a complete behavioral/demographic profile?
We don’t see all users all the time, but through our publisher and marketer partners we are able to interpret the media consumption of virtually all US Internet users when they interact with their content. We analyze a very broad range of activity – over 6 billion consumption events every day – and this comprehensive anonymous view of Internet media consumption enables us to model the web with a scale and precision that has never been achieved before.
In terms of being ubiquitous, it’s less than three years since we launched and we think we’re on the right track. Already more than half of the top online media companies (as ranked by ad revenue) have joined our service, and the adoption rate among top marketers in the three months since Quantcast Marketer launched has been phenomenal. So long as we continue to focus on delivering value to our customers we’re confident that over time the vast majority of ad-supported sites and networks will participate in our program.
Where does placement fit in your strategy? It seems like it’s all about audience. Is it?
Our strategy is to provide the marketplace with a set of services that enable ad sellers and buyers to define and control how they transact. While sponsorship and property-based ad models will continue to be highly effective for many advertisers and publishers we believe these have to be complemented by addressable audience models that allow marketers to define the types of consumers they want to reach, and deliver them at scale across the web. In fact, we expect to see placement and audience value coexist as driving forces in the media buying and delivery process, often used together (i.e., a marketer just wants to deliver men, in top tier professional sports content).
In its most basic form, placement has been the proxy the advertising industry has used to deliver audiences since its birth. An advertiser bought a TV commercial in a particular show or page in a specific magazine, because its audience tended to have a certain set of characteristics. There is waste in property-based models – not everyone consuming a particular media property shares the same attributes. Even though sports content tends to attract men at a higher than average rate than women, there are still women consuming sports content. Ultimately, what an advertiser wants is the ability to minimize waste – and deliver the highest possible concentration of the audience they care about. In the analog world, the reason advertisers buy properties and placements is because it’s their only choice – media is delivered on a one to many basis. That changes in a digital world, where every media consumption event is served in real-time.
Follow Quantcast (@quantcast) and AdExchanger.com (@adexchanger) on Twitter.
