Media Dojo

Archive for the ‘New Advertising Models’ Category

It’s my fourth year going to Monte Carlo for the Monaco Media Forum. The MMF is an invitation-only get together of about 350 digierati from the media and marketing worlds. Groupe Publicis (the number 4 advertising holding company but number 1 in digital revenue) and HSH Prince Albert II are the main patrons, along with Microsoft, Lenovo, Orange, Google (YouTube) and Booz & Co. I’ve helped with the agenda for a couple of years now. There will be a special panel on media clouds that will include Media Dojo alum Sean Knapp from Ooyala among others. At the event, I’ll launch the MD Guide to Cloud Computing for Media and Marketing. The Guide will be a first stop for business layer media and marketing professionals who need to get a handle on the foundation concepts, tech, companies, and impact of cloud computing on their industry.

If you’re interested in attending, drop a line at

contact@monacomediaforum.org

to request an invite. If accepted, you’ll pay for air travel and hotel with all the ground costs picked up by the organizers.

Online video is huge in terms of users. Online video is more huge in terms of usage.

So where’s the money?

That seems to be a standard story line these days in both general press and on blogs. Everyone spouts off about whether online video will overtake TV, when and how. The inserted bias to these stories is that online video is a mathematical function of television, as conceived and organized by the television industry. By definition, if the image moves or is animated, it must be either film or television. But that’s a lot like benchmarking an Indy car against horse drawn carriages because both have wheels, need a road or a track, and are used by people for transportation. You can figure where the logic is leading….

So, it was a breath of fresh air for me to speak again with Sean Knapp, co-founder and CTO of Ooyala regarding a new project the video platform service provider is doing with Wheels TV and eBay Motors. For more corporate info on Ooyala, here’s the Media Dojo Tear Sheet–ooyala.

Ooyala is working with Wheels TV to market test POV (pre-owned vehicle) reviews on eBay Motors.  It’s a new, video-based consumer shopping service for people looking for pre-owned vehicles. Each five minute POV video review contains road tests, walk-arounds and data addressing reliability, safety and fuel economy information related to about 200 make/model/year automobiles for sale on the eBay Motors site.  The POV Reviews are produced in cooperation with J. D. Power and Associates.  JDPower.com’s Power Circle ratings suggest trends in overall dependability, performance and quality on every vehicle. POV reviews also include mileage estimates from the Environmental Protection Agency and crash test videos (yeah buddy!!!) from both the Insurance Institute for Highway Safety and the National Highway Transportation Safety Administration.

Naturally, viewers can share the videos across their personal networks.

Here’s some sample videos:

BMW 3 Series

Ford Mustang

Honda Pilot

The primary benefit to the buyer is that video can crunch several hours of research on the make/model of a given automobile into about five minutes. Translated, the videos give the buyer quick, effective talking points for persuading their significant other around the dinner table that it’s the JD Power *safety* rating that makes the BMW 3 Series a smart buy, rather than the kick-ass pick-up and handling, not to mention the fire-engine red color and awesome trim.

I spoke with Sean about how Ooyala is handling the video demand, especially from the viewpoint of analytics.

Media Dojo: What kind of analytics will eBay Motors get with these video streams?

Sean Knapp: They’re basically getting the full suite from us ranging from geographical breakdown, to the unique user base to how many uniques they’re getting on a daily, weekly and monthly basis. They’re also getting behavioral analytics that pinpoint which particular part of the video people watch, what’s the abandonment rate, who’s skipping ahead. Then, they can start looking across video to compare the acquisition/retention curves for the Ford Mustang versus the GMC Envoy on the site. Finally, eBay Motors is using our API to pull in data that they’ll crunch using their proprietary in-house analytics systems.

MD: Are online video analytics going the way of other digital analytics in becoming more performance-based as opposed to just exposure-based?

SK: How we monetize content isn’t based as much on the number of impressions anymore. The issue is that over the next few years there’s going to be a 10-30X increase in overall video content served even though there will be only a 2-3X increase  in the number of viewers. So what users are doing with video becomes the key metric to track as opposed to just who is being exposed to video. What video content are users accessing? How are they consuming that content? How are they responding to advertising?

MD: Granted the need for better analytics, are publishers really becoming more sophisticated about using video?

SK: On average, people are getting more sophisticated on the buy side. Publishers are looking more closely not just at how consumers are consuming their content but how the publishers are monetizing that content. Over the course of the past six to nine months, we’ve gone from supporting  2 or 3 ad networks to supporting 12-15 different ad servers and ad networks plugged in to our platform. Publishers are getting away from saying just “How do I get video into my site” more to “How do I refine the video on my site? Which knobs should I turn to get people to consume it? How do I extract value from that consumption?”

MD: Obviously, eBay Motors has a clean benchmark for monetizing the content (e.g. brokering sales). What about other sites that aren’t squarely in the e-commerce bullseye? What trends are you noticing in terms of their ability to monetize video content?

SK: In terms of monetization, there’s no silver bullet. There’s some value in a video CPM and some value in a CPC. But it all eventually falls under the umbrella of some kind of Cost Per Action. We think the better players will be those who carve up a broader publishing base into finely sliced niches against which people can advertise. Auto is a good category in which there are numerous niches for targeting that can be aggregated into a bigger media buy. But to get to that place, you’re going to need to see the larger video platforms get into closer collaboration with the larger ad networks. Everyone needs to help create a larger media buy ecosystem. To get the best exposure, brands can’t just dip into the top 100 sites of a given category, but need to get into the top 100,000 sites. This means that a lot of mid tier publishers using video will need to offer more sophisticated analytics to get that business but it’s not likely that they’re able to build that in house. That’s where the large video platforms like us come along.

MD: Last question. How much cloud computing horsepower have you added to keep up with demand since we spoke last spring?

SK: We’re seeing anywhere from 30-40% growth on the low end per month to over 100% growth in certain months. It depends on the metric you chose whether it’s GB ingested, video hours served, video users reached. We’ve been able to scale things through good partnerships with our Content Delivery Network (CDN). We also have a very good distributed computing team in house. We built our transcoding and storage applications to site on top of cloud infrastructure from day one. Today when you upload a 2hr length full movie to us, it will hit anywhere from 10-100 different encoding machines operating in parallel. We can now encode a HD quality 2hr movies substantially faster than real time by operating in parallel on cloud infrastructure.

I’m In Langley, WA this coming Saturday September 19th to speak about the media play for cloud computing at New Media 2012. I like the agenda and set up. Each speaker gets five min to make their case. Then comes a panel discussion. The line-up includes people from the telecom world, gaming, visual media and journalism. Here’s the speaker list:

Tom Kennedy, Former Director, Multimedia, WashingtonPost.com
Brent Friedman, Partner, Electric Farm Entertainment
John du Pre Gauntt, Author, Consultant, Technologist
Joe Pulizzi, Junta 42; Author, Get Content, Get Customer
Alexis Gerard, Founder, Future Image Report
Robert Gilman, Founder, Context Institute
Russell Sparkman, Founder, Fusionspark Media
Marcia Hofmann, Staff Attorney, Electronic Frontier Foundation
George Henny, Co-CEO, Whidbey Telecom and Fibercloud
Joseph M. Tringali, Co-Founder, General Manager 5TH Cell Media

There’s also the venue—the Clyde Theater. It’ll seat about 200. Tickets are still available for the event which will run from 1pm until 330pm Saturday.

Hope to see you there.

Over 100 million games installed in 18 months. 40 million monthly players. 9 million daily players. Already profitable. 200+ staff spread around offices in Europe, US and China.

And, by the way, they don’t own a single server.

It’s often the case that when you use the terms “social”  or “cloud-based” to describe a company or a business venture, a lot of people roll their eyes—often for good reason. The bullshit to hard numbers ratio often exposes that you’re dealing with just the marketing phrase.

In the case of Playfish, the numbers speak for themselves. Founded in late 2007 by people with deep roots in the mobile ecosystem (e.g. Nokia, Glu Mobile), Playfish launched to take advantage of Facebook and MySpace as distribution channels for games. For quick overview of the company, check out the MD Tear Sheet.

Earlier this month I spoke with Sebastien de Halleux, Playfish co-founder and COO, as part of a research project for GigaOm Pro. The following is a more extended interview.

Media Dojo: Please define social gaming…

Sebastien de Halleux: Social gaming focuses the value on the interaction between friends via games as opposed to concentrating the value on the product side by selling a copy of a game. So we’re heavy on the idea of a game as a service and the connected nature of social experience as opposed to concentrating the value on the product side of selling a copy of a game.

MD: To state the obvious, then, how do you make money?

SH: From a business model point-of-view, the Playfish model employs micro-transactions during the game as well as in-game advertising.

MD: How does Playfish operate in the cloud?

SH: All of our business and commercial infrastructure runs on the cloud. The company literally only has laptops. We use Amazon S3 for storage, EC2 for computing, and use Cloud Front for content distribution across the fixed and mobile web. We were founded in October 2007, which was the first time you could get an entire company purely in the cloud.

MD: There’s a lot of mobile DNA in the company as well, how do you see social gaming in the mobile space?

SH: Today, there are about 1.5 billion web users worldwide who have access via a laptop or PC versus 3.5 billion mobile users, many of whom have some level of access to the Internet on their mobile device. So there’s already a very strong skew toward mobile computing in general as an access layer, especially outside the US. Social gaming is all about bringing a shared experience to a group of friends, wherever they are, whatever their access preferences might be.

It’s clear that more people are shifting a greater amount of their time from the bigger screen stationary experience to the smaller screen mobile experience via smartphones and netbooks. It’s not a function of us dictating that the user must have a specific device to enjoy a particular kind of content experience. It’s a function of the user choosing the stream that’s most adapted to their lifestyle at any given moment. As a service-based game company, you need to be able to offer a meaningful experience to people regardless of their access device or the social gaming model runs into problems.

MD: How does this work in practice?

SH: We launched an iPhone and iPod Touch version of one of our most popular titles, “Who has the Biggest Brain?” at South by Southwest this past March. It reached number 2 position in the Apple App store in the UK and was a top 5 mobile game in many other countries. The look and feel of the game is the same whether you play on your iPhone or your laptop even though the underlying rendering technology and input method is different. The key similarity is that your friends, who are Facebook friends in this case, are present as part of the experience regardless of whether you’re using your iPhone or a laptop. Moreover, your score and your progress within the game are preserved even if you pause it, change access devices, and re-start.

What’s important for the user is that they no longer think of the device as the platform but much more as the means to connect to the service. For us, the “platform” is the social network.

MD: What needs improving for the user to forget or not care about the access device they use to play one of your games?

SH: The iPhone is a revolutionary mobile media device but has atrocious connectivity once you get away from a Wi-Fi connection. It’s typically a use case never to assume 100% connectivity to a mobile cloud service. Managing this duality of being based in the cloud, but needing to maintain availability to the end user on questionable infrastructure is a huge technical and design challenge. Practically, this means that we still need to have the user download a client application into their device, mostly to handle the lack of bandwidth plus the frequent interruption of mobile connectivity. You often need to push updates to the client in order to update the service, which isn’t a good experience. So one of the milestones we want to see is the day when you can run a game in a mobile experience using Flash or something comparable to the web development model instead of needing a specific mobile client. That will require a big jump in mobile connectivity.

MD: What about on the cloud computing side? What needs to happen to take it to the next level?

SH: In terms of the cloud industry ecosystem, there are several forces in play. First, we are reaching a scale with 40 million monthly accesses to where we start to have increasing leverage as buyers of cloud services. When we started, AWS was the only credible provider. They’re still a close partner. But I think the market needs more mature options among different cloud providers, especially across geographies. At the moment, we’re very happy with our current partner but I think it’s important for the cloud industry to stay competitive. The second thing is more cloud specific services for businesses like ours. For example, our business depends heavily on micro-transactions. But it’s not uncommon for certain online payment providers to require a static IP address, which is a contradiction in a cloud environment. How to design a cloud-native billing layer is still pretty open territory. The next one would be cloud enabling layers, the tools that enable a company like our to manage and optimize the operational aspects of running a cloud-based business. Sure, there are companies focusing on this problem but many are tied to existing cloud providers. We’re like to see more cloud agnostic tool providers.

MD: Last question, how do you see games pushing cloud computing forward?

SH: Games will help cloud computing progress because of the sheer scale of demand it puts on the infrastructure. We’re probably one of the largest AWS Cloud Front customers because we need to push huge Flash files all over the Internet. This is a demand level you don’t see as much with business applications. The demand for game services will play a big role in developing cloud infrastructure, and especially tailoring it for low latency, efficient distribution and always-on connectivity. Latency and performance is everything if you’re offering games for multi-player experience.

To date, my idea of a “serious game” revolved around a pool table, money and a shot of straight liquor. PC-based or web-based games seemed more appropriate for the kids or geeks. It’s stereotypical, I know, but supported by years of anecdotal evidence.

However, I’m changing my view of games after a trip over to Breakaway Ltd. in Hunt Valley Maryland. At the end of May, the Knight Center for Specialized Journalism bussed us over there to hear about serious games. We learned that the Pentagon as well as healthcare institutions like Texas A&M took games very serious indeed. For a quick one pager on Breakaway, check out the Media Dojo Tear Sheet—Breakaway Ltd.

The problem is thus: how do you enable people to tinker with real concepts based on real situations—but without real consequences? Just as product engineers take great pains to stress their inventions to the breaking point before deploying them in the field, many of those who play for the highest stakes (war fighters and healthcare) are looking hard at game-based simulations to help them test-drive strategies or ideas before they go live.

I followed up with Breakaway CEO Doug Whatley this week to discuss how serious games might impact the cloud if, as we both agree, it’s a category that’s likely to diffuse beyond the defense and healthcare communities to other industry sectors.

Media Dojo: First, tell me how games chew up computing resources

Doug Whatley: Part of the reason that games are so cutting edge is that people often need to buy the latest and greatest machines to play the latest and greatest video games. So the best games always have a very high machine spec compared to the installed base. Since serious games are coming out of modeling and simulation, they fit naturally for the military because they were already training people in flight simulators or something similar. And for medicine, the fact that they have this very expensive, sophisticated dummy attached to a PC meant that medical customers were accustomed to buying a lot of equipment as well. The issue now for serious games is that if they’re to move beyond military or healthcare to become a regular part of training and education, the client hardware issue grows in importance. Often, you don’t find anything other than really old machines throughout government and the civil sectors. If you go into your local fire station, they’re likely to have a really old PC. So the ability for us to deliver them the latest training (or not) via a game simulation is more of an issue of them not having the equipment to run the software.

MD: Does Cloud Computing solve this problem?

DW: The appeal about cloud computing is that someone can sign on over the web and play the latest training game in the cloud without having to upgrade their local PC. That’s the perfect world. But in all areas of storage, processing and bandwidth, the resources in the cloud are being pulled pretty hard for games. The amount of data necessary for a lot of these simulations can be very large. Whether it’s the military or Homeland Security with their terrain databases, all of the DBs backing up that information are very large and all that data needs to be stored somewhere. If you have satellite images and other earth sensing data, it adds up pretty quick. To show people details of an actual city, that’s many GB of data. So the data storage and having a centralized location to keep that up to date is very useful. But because there’s so much data the throughput is very important as well. And the fact that you’re doing very sophisticated physics based simulations against those environments means that the processing is really getting hammered.

MD: What are the holes that need to be filled by cloud providers to drive adoption by game publishers like Breakaway?

DW: One of the biggest holes for adoption of cloud by games makers is the virtualization of certain kinds of hardware like 3D accelerators. It goes back to the processing needed in the cloud to where the web-based user experience mimics what you’d find if they had a latest version machine. There’s just not enough availability of virtualized hardware based in the cloud to really make game-based simulation a broad-based offering.

Another thing is that right now the cloud service providers are focused on the e-retailer market rather than the application vendor or Service Oriented Architecture (SOA) market. Cloud providers need to step it up on the app vendor side as opposed to just expanding my sales capacity. Every cloud vendor has a riff talking about your web site getting mentioned on Good Morning America followed by a huge traffic spike. That variable burst capacity is great. But what if I’m an app service vendor and I’m pushing your cloud that hard all the time? It’s a different situation.

MD: Let’s imagine the cloud vendors are able to virtualize hardware better and provide an app-centric as opposed to website-centric service offer. How would that change your business of using game technology for simulations, modeling and training?

DW: What cloud-based delivery enables me to do is change my business to a subscription model. For example, if I create a fire training application, and allow firemen throughout the world to subscribe to my training service, that gives me such global reach that it completely changes my business economics. Right now, I’m out there trying to sell applications in a box and get people to buy them and go install them. With a cloud-based approach, I can switch to a more subscription based model. That’s actually what my customers want. That’s how the training budgets are paid for now. This goes back to the modeling and simulation versus training. It’s very common for companies or government employees to say, “I’m willing to pay $9.95 per person per month to have all my people trained in all these different areas”. That’s how they’re naturally inclined to purchase versus now where I have to convince them to buy in a different way.

Time for the second installment of last week’s Amazon Web Services (AWS) meeting for local start-ups in Seattle. Rounding out the four customers presenting last Thursday were two Seattle-based media plays, nuTsie and Zumobi.

First up was Bob Wise, VP of Engineering of nuTsie, (www.nutsie.com), which allows people to port their iTunes play list across web and mobile platforms, including Blackberry and iPhone.  Basically, nuTsie takes a user’s existing iTunes library and rolls it into a streaming service much like Pandora. They don’t use the actual music in the library but the meta data about the songs and/or a playlist to create a super customized experience anytime, anywhere. If it seems a little disjointed there is method to the madness. Music licensing remains a mess even after a decade of industry tinkering. Like Pandora, Melodeo must make all its music streaming DCMA compliant so legally nuTsie is considering web radio rather than a a formal music distribution service.  The primary outlets are streaming for the web and for mobile phones. The business model is based on advertising for web streaming and subscriptions mobile phones.

For plumbing, Melodeo uses Amazon S3 to store and serve up the audio files (several TB in aggregate) that stream via a Flash player. The web-based nuTsie service gets about 10 million page views per month with about 10,000 hours of streaming music content served up each day between the web and mobile components. Both the streaming service and the mobile play are hosted on AWS. Bob said that for a typical load, it takes about 40 EC2 instances (think 40 virtual servers) that are about evenly split between large and small instances with one extra large instance for the main database. If you do a back of the envelope calculation it works out to roughly $10-15 per hour for pure compute capacity. Remember that nuTsie is also paying for data and certain transfer bandwidth charges.

source: http://aws.amazon.com/ec2/#pricing

One aspect of Bob’s presentation I liked was how he illustrated the effect of business forces on technical design. Chris Anderson of Wired fame used music as exhibit A of his Long Tail hypothesis. Bob said that in his experience the long tail might be long but it’s also thin as fishing line. Basically, this means that ultimately the number of music plays instead of the number of music tracks is what makes or breaks the business. Given the fact that the action stays with a relatively small number of tracks, nuTsie uses Amazon S3 as a content delivery network (CDN). If it sounds strange to use a data storage service to serve up content, take a look at charging. With many other CDNs in the market, a business is charged according to how much data sits at the edge node plus the transfer bandwidth to the end user. Thus, the key cost point is how much you get charged for keeping music tracks in storage which aren’t being played very much. Sticking several TB of music data out there on various edge nodes is an expensive way to do things. If you look at parking data similar to parking cars, loading rarely played music or video on an edge node is a bit like using a parking meter or a temporary lot whereas oft-played content needs the equivalent of a monthly reserved space. It’s an imperfect comparison I know. However, it’s decently clear that some of the heavy lifting for media providers is to figure how thick is the head of their demand model and how thin is the tail. Otherwise, it’s money out the door, cloud or not.

Ask people in the abstract whether they want more choices,the answer is typically a resounding yes. However, look at how most choices are presented in real time and most people end up deciding between black or white, good or bad, open or closed, Democrat or Republican, whether they’ll have the chicken or the fish.

Digital privacy policy is often set up in a similar vein. You either have it (privacy) or not.

Binary choices can be useful for setting up the infrastructure for competing views to join in battle. The contestants line up in their respective colors. They go to various media outlets as either the home team or the visitors. There are lusty cheers or jeers greeting their positions, and we’re told that through this kabuki the winning view of truth will emerge.

But a funny thing happened on the way to Capitol Hill with digital privacy. Try as we might, it’s been almost impossible to set it up as a simple choice.

When I first applied to the Knight Foundation fellowship, I naively thought finally, I’ll understand digital privacy once and for all. That was laziness on my part. But, hey, imagine if it actually panned out. I’d have been a pretty smart duck.

The closer truth is that after hearing some very talented and passionate speakers along with sharp questions from 23 journalists, I’m believing there won’t be a definitive ceremony marking either the end of privacy or the complete securing of our privacy rights as they apply to the digital space.

More likely, we’re in for years of ebb and flow of digital privacy issues on the policy radar screen. Digital privacy can’t be solved as a specific problem. More likely, it’ll be managed as an ongoing issue with flare ups and periods of quiescence. I believe this isn’t simply because we haven’t sorted through the batting order of privacy contestants. More close in my opinion is the reality that you can’t separate the practical debate (how we do it) from the philosophical debate (what are we protecting) from the political debate (who decides it).  It’s three level chess.

We’re building an economy in which demographic and activity data is becoming the dominant form of capital. By dominant, I mean that it leads the dance for other forms of capital. In the 18th century the valuable thing to own was land. I could use my title to land to take out a loan to purchase something else. In the 19th century it was about owning machinery. In the 20th century, I’d say it was about owning formal intellectual property although I freely admit I’m agnostic whether that was the definitive form.

2009 is a bit more clear to me. With real or de facto ownership of personally identifiable information combined with analysis of a stream of activities, I’ve got a pretty good idea of a person’s economic impact and potential future value. If I can encapsulate that knowledge into some digital artifact I can transfer to someone else (e.g. a customer record, search history, ticked preference boxes) then that’s property. I can use such property to organize other forms of capital such as financial, technical, human and so forth.

Twelve years ago, I gave a talk at Harvard’s JFK school about the political economy of the Internet. It was 1997 and everyone had policy theories.  At the time, I tried to fit Internet policy debate in the context of property rights as I understood them from grad school.

“…the new reality is that the creation of wealth and power is shifting toward applying intellect, technology, and economies of scope to the problem of production and exchange as opposed to energy, labor, and economies of scale. No longer is the individual firm the fundamental economic unit but networks of firms where the integration of knowledge is superceding the division of labor to define an economic system.

Is this the new economy as trumpeted by the media? Perhaps it is, though the data is very incomplete at this point. But one thing is clear. We are witnessing the early days of a struggle between intellectual property and classic financial markets to define capitalism’s center of gravity. We can expect a wild ride for the next few years as Wall Street tries to value intangible digital assets with intangible digital securities. They’ll eventually get the trick right but history suggests that more fortunes will be lost than made before the dust settles and we have a new set of values.

Moreover, it is apparent that the Internet is playing a critical role in this struggle because it is becoming the main institution for circulating and adding value to intellectual assets or claims on those assets in the same fashion that the banking system circulates financial asset and claims upon them. This is changing the way that credit is created, bought and sold and therefore, the way in which the use of capital is determined, and how people are organized for work.

And those individuals who internalize this distinction and improve upon the process will control significant wealth and power in the 21st century.”

Well, it’s the 21st century and I was wrong about a major part of that analysis. I thought that intellectual assets meant intellectual property (e.g. patents, copyright, trade secrets, trademarks, brands and logos). What’s happened instead is that mindshare, usage, registrations, searches, in other words information about people has emerged as the most important asset for a new economy.

Digital privacy policy can’t be restricted to economics, politics, law, culture or technology. It’s all of that because we the people, you and I the individuals, are the primary assets in play for a new form of capitalism. In that sense, I wouldn’t be surprised in my lifetime if we end up amending the US Constitution to accomodate the new realities.

Publishers have an inventory and supply chain management problem similar to airlines and their seat pricing. The value of the asset changes rapidly, changes according to who wants it, changes according to how it’s bundled with other offers. It’s a whirl of activity right up to the point when the plane takes off or the ad impression is served whereupon the value of the asset plunges to zero. It’s an everyday challenge to practice effective yield management for both online ad spaces and airline seats.

Savvy operators in both industries have long known what they wanted to do to crack this problem: take raw data in the form of server logs or ticket price searches, refine it into information that can be combined with other information in an analytics package, process that information into insight about opportunities to maximize revenue at the margin, and then take action based on that insight. It’s a relatively straight forward process that, technically speaking, requires a shitload of data, storage, processing, and reporting infrastructure. Thus, the “cost-effective” constraint raises its ugly head again.

However, cloud computing is starting to change that dynamic. I spoke with Tom Shields, CEO of YieldEx which is a cloud-based provider of tools for publishers to manage and optimize their online ad inventory, including forecasting of overlapping inventory and best possible campaign allocations. YieldEx became the second company to win Amazon’s Web Services Start-up Challenge, receiving $50,000 in cash, $50,000 in AWS credits and eventual direct investment into the company. For a quick overview, access the MD Tear Sheet. MD-TearSheet-YieldEx

Media Dojo: Why the focus on cloud computing?

Tom Shields: We ran into a number of problems during the 90s and into 2000 that were essentially unsolvable with the compute resources and capabilities we had at the time. The volumes of data we needed to process, the types of analytics that people wanted to do, the kinds of information that people were trying to get out of that data was pretty hard obtain cost-effectively. But we knew what we wanted to do. It’s a combination of yield management and plain old supply chain management for publishing. So what YieldEx is trying to do is apply computer science to and new capabilities offered by cloud computing to advance the state of the art.

MD: How are ad spaces different from airline seats when it comes to yield management?

TS: There’s a couple of things that make advertising impressions unique. Like airline tickets they disappear and the value goes to zero. If you have a physical object on a shelf, it may decline in value over time but it doesn’t disappear. The second interesting thing that so far seems unique to advertising impressions is that impressions can be characterized in many different ways. Those different ways have different values to different people.

So an impression coming in may be a male 25-34 on west coast looking at sports content. You might have some marketer wanting to buy sports content opportunities, others want to buy males; still others want to buy west coast. And these marketers may have different values attached to each one of those things. Part of the challenge here is that what one person considers premium inventory, another considers junk, what works for one marketer might not work for another. So there’s an optimization challenge to figure out how do I as a publisher maximize my revenue given this set of X impressions that expire at a given Y rate and given Z demand from all these different marketer audiences. It ends up being very complex and challenging, especially when you have 10 million ad impressions being served every month and you’ve got 10 thousand potential ad placements. The cloud allows us to economically crunch that kind of data unlike before.

MD: How does this work in practice?

TS: We take in a day’s worth of our publishing clients’ server log files. We process those log files overnight, turning them into clean information sets that you can run queries against in order to surface an opportunity to optimize your ad inventory. We need to be able to start up a hundred instances of significant sized virtual machines, and run them all in parallel for a relatively short period of time like an hour, 30min even. Then we need to shut them all down again and stop paying for them. That’s why our application is particularly well suited to the cloud.

In terms of the amount of server log data we need to process, it ranges from a couple of GB to a couple of hundred GB per customer per night. For some of the larger publishers, their current reporting infrastructure takes them 22 hours to process one day’s worth of server log data into information they can sell against. If you take that same data and put it in a cloud based parallel infrastructure, you can process it sometimes in less than an hour. It’s a huge difference.

MD: Is the nature of impression data changing online? How do you optimize that?

TS: Along with the Facebooks and other social platforms generating mountains of impression data, publishers are going to create a lot of that stuff anyway. How they obtain that data and use it for their own benefit and potentially sell data as well as impressions is probably where we’re going over the next 3-5 years. Publishers will probably also be combining their impression data with 3rd party impression data to get better demographics so that they can generate profile information that they can turn around and sell more effectively than an ad network might be able to.

Then there’s going to be this whole concept of measuring performance. What’s interesting is that for a number of marketing objectives that brand guys typically talk about, clicks are really a terrible measure for those things. For the publishers, the trick going to be about attribution tracking. It’s knowing, “sure that guy went on Google and did a search on the product but he saw five banner ads for that same thing before he searched and wouldn’t have known what to search for without getting the awareness from those display ads”. How does a publisher get credit for that? He won’t unless he can prove his content had an impact. And that requires data.

MD: Why do clicks rule? Should they rule for branding?

TS: It’s easy to optimize stuff you can measure. If you look at optimizing around clicks, it’s actually pretty easy because you can measure the clicks right away. You can promote your site because you can see where people are clicking and where they’re not. Without minimizing the scaling challenges involved, it’s a pretty straight forward problem. If you have feedback, you can optimize rather quickly.

The problem is that it’s hard to get feedback from activities that move higher in the funnel. Take brand awareness. How do I get feedback on brand awareness I can quantitatively measure automatically in software? If I can’t measure it quantitatively, it will be much more difficult to modify the format or delivery of the ad to optimize it quickly. If you can start measuring those things, you can start optimizing them very quick. But part of the problem is that there is this time delay between measurement and feedback because you can’t measure the impact of brand awareness until much later.

MD: Last question, do you see cloud computing as a means to collapse this float between measurement, however defined, and the ability to take action?

TS: Cloud computing offers only the ability to ingest, process and report a lot of data under far better economics. It’s no replacement for common sense on the publisher side. Whether you use cloud systems or what ever the key to optimizing your inventory is to learn what works with the customer, then go after it and earn money. How fast can you run your learn cycle so that you can maximize your earn cycle.

For marketers, learning is the overhead in the overall media buy. You’re taking your hit via lower effectiveness during the learning period of the campaign cycle in order to maximize the effectiveness of the earning period. So the shorter and more effective you can make the learning period, the less overhead you have, the more efficient the overall media buy becomes.

Here’s the first of a series of presentations I’m working on. I’ve loaded a 9min screencast of a paper called “The Retail Future of Media and Advertising’. In the presentation, I contend that in a world of near infinite digital shelf space, successful media and advertising strategy will look more like consumer retail. It’s a think piece. I will follow up over the next few weeks with some data-centric presentations to back up some of the claims made here.

If you prefer reading, here is a PDF written in 2006 that contains the main argument.

The Future of Media is Retail

The last couple of interviews I’ve done with ooyala and Tumri have me thinking that we’re getting closer to a Just-in-Time World for media and marketing. By that, I mean that the action (the ability to profit disproportionately) is moving to how fast and how well you can customize a content experience or an ad at the margin. It used to be that leverage was all about aggregating eyeballs. No argument there in terms of creating an efficient media buy for a big advertiser. Scale has its advantages and always will.

But as we’ve found in other industries like automotive or consumer electronics, scale only makes you a player. Soon enough, you get a disruptive force like Toyota saying that you can customize the hell out of your car and we’ll build it, and ship it to your dealer in a week or less. That’s where scale takes on a whole new meaning. It’s no longer only the ability to spit out more units at a lower per unit cost. Scale becomes a game where the trick is how fast you can orchestrate a massive ecosystem to deliver an individualized experience.

I’m not saying we’re there today with media and advertising. But some of the building blocks have been put in place. Cloud infrastructures are becoming the assembly lines for a new breed of media and marketing disruptor. Whether it’s Tumri blowing up display ads into objects that are then re-assembled on the fly based on impression data. Or ooyala seeding video streams with a slew of clickable elements, the game is shifting. Like it or not, principles of e-commerce are being injected into media and advertising just like retail, travel, real estate, automotive, and financial services.

The advantages of scale will never go away. But how those advantages get re-distributed along the value chain and which new players control them, ah, now there’s an interesting question.

Stay tuned over the next couple of weeks. I’m banging out a 5-6min screencast called “When Media Acts like Software” to take a first stab at the idea of a JIT world for media and advertising. It’ll be a first rev so it’ll eventually join my 8th grade class picture at the bottom of the Marianas Trench. But that’s often the only way to learn. Just F*ckin’ Doo It as Nike would say had it started in New Jersey.