Oxyfish

This is our third year at the Crunchies, and I think this could be our best effort yet. Kudos to Matt Hempey and Mark Casey and the team that put this together – I basically showed up and sang. And danced. A little.

Google’s release of the DoubleClick Ad Exchange 2.0 has introduced Real-Time Bidding (RTB) to a much wider audience. While they were not the first, they are probably the biggest, and their entry is starting to legitimize RTB as more than just a niche.

Neal Mohan’s introductory blog post emphasizes the three main principles behind the development of their exchange: simplify the system for buying and selling, deliver better performance, and open up the ecosystem. It’s this last point – openness – that I’d like to explore.

Real-time bidding offers some openness for the buyers: they are delivered each impression, with the floor price, URL, and cookie, and have a fixed amount of time to bid. They are then notified if they win and a request is made to deliver the advertisement. What’s surprising is that unlike a standard auction, at eBay for example, if they lose, the potential buyer has no idea what the winning bid was. Google gets to keep all that information.

Even more incredible is the fact the publishers also aren’t told the winning bid amount. They get an aggregate value for their earnings, but can’t see the value of each impression. This is as if you auctioned 10 things on eBay, and at the end, eBay sent you $100, but refused to tell you what item 1 sold for vs item 2 or item 3.

This information asymmetry is largely to the benefit of Google, but also skews to the buyers. Savvy buying systems can tweak bids up and down in real-time to do crude discovery of the “true” value of different kinds of inventory and how it varies over time. Publishers have no such ability to discover their inventory value at an impression level. Worse yet, while buyers can bid different prices for each impression, publishers have no ability to re-set floor values on each impression to push the bids up. Of course, they would need new tools to do this (SSP, anyone?), but it is much harder without data.

One final point on how the system is stacked against the publishers: any buyer can participate in any of the exchanges, and indeed many of them do. But since Google does not give publishers their impression values, it is very hard for publishers to find out if some of their inventory would perform better on a different exchange. And to add insult to injury, Google makes it almost impossible for non-DFP publishers to participate at all.

Publishers should be wary of using any ad exchange until they get real openness, and the tools – like an SSP – they need to ensure the deck isn’t stacked against them.

This post also appears at AdExchanger.com in “The Sell-Sider,” a column written by the sell-side of the digital media community.

I’m getting to the age where if I go play a sport I haven’t played in a while, I ache for days afterwards. I recently played 5-5 full-court basketball for the first time in decades, and could barely walk the next day. Some of those hurts are “good hurts” – sore muscles that are getting stronger from the workout. Some are “bad hurts” – like a partially torn rotator cuff. They all hurt, but it can be important to distinguish between them, because the remedies are different. And no matter what you call them, they still hurt like hell.

Similarly, in startups, we talk about “good problems” and “bad problems”. Bad problems are the ones nobody wants: unhappy customers, products that don’t work, or markets that don’t materialize. Good problems are ones that sure seem like they’d be nice to have: too many customers sign up at once, investors want to put in too much money, etc. Just like bad hurts and good hurts, bad problems generally require outside intervention to fix, while good problems work themselves out through positive progress. Everybody says they want the good problems, but they are still problems – and they still require a hell of a lot of work to get through.

The new real-time bidding (RTB) exchanges seem to skew the buying power further in favor of buyers. They can see each impression in real-time, data-enhance it with their own data, and then bid on it. The problem is that most publishers don’t know where the pockets of value exist in their remnant inventory, so they can’t intelligently allocate that inventory in a way that makes them the most revenue.

Publishers need 3 things to maximize the value of their inventory on an exchange:

1. Inventory value. Publishers need to get back from their exchanges the value of their inventory, on an impression-by-impression basis. Just getting high-level average CPMs by section or zone isn’t good enough, this masks the high-value impressions that may exist within those sections.
2. Third-party data. Your buyers are using data to understand the value of your inventory, you need to have the data to fight back. You should be setting floors on your inventory based these data segments, so buyers can’t just cherry-pick your inventory at an overall low value, and you can’t do this without the data.
3. An analytic system to maximize yield. You will need a system that can capture an analyze all this data – terabytes of it – and spit out a set of floor prices by inventory segment. In an ideal world, this system operates in real-time, re-setting floors based on the most recent data. I call this real-time selling – the antidote to real-time bidding.

Not all of these items are easily attainable. The first item is not generally available from exchanges, so publishers need to demand it. The second is becoming more available, from vendors such as Audience Science, BlueKai, and eXelate. The third will be provided by Yieldex, among others. Forward thinking publishers, who put this stack together, should be able to dramatically increase revenue from their unsold inventory.

This article was also published on AdExchanger.com.

Ever run across a Windows app you really wish you could run on your iPhone? I did, just the other day, planning a 1-day kamikaze trip to Disneyland with my 3 kids. There’s an excellent program called RideMax, that helps optimize your day to minimize wait times and hit all the best rides. The idea is to make a plan, print it, and more or less stick to it. Clearly the developers don’t have kids – no plan survives 3 kids for long. What I really wanted was the ability to run it again a couple times during the day with the remaining rides we want, and get a new plan each time. But that would mean running it on my iPhone.

You can’t really run Windows applications on your iPhone without cheating somehow, so I cheated. I loaded the free LogMeIn client to my desktop PC at home, and bought the $30 LogMeIn Ignition iPhone app. This worked amazingly well – I was able to log in to my desktop and control RideMax just as if I was sitting there. And it was especially great when I ran it while waiting in line at the Alice in Wonderland ride (which was a last-minute addition by my 4-year-old) to figure out if we should do Splash Mountain or Space Mountain next. I’m confident the program saved us at least an hour of line time, and re-running it during the day was a key factor.

One complicating factor for me is my iPhone is a 2G and I’ve canceled the AT&T service so it’s essentially a first gen iPod Touch. To get online, I used the nifty Verizon Mifi that I have for my laptop. The technique is described in more detail in this Verizon iPhone article. So I was running LogMeIn on the iPhone to control RideMax on my home desktop, connected via wifi to the Mifi, which connected via the Verizon 3G network and the internet to my home DSL, and through my internal network to my desktop. There were lots of moving pieces, but in this case they all actually worked together nicely, and the result was a great day at Disneyland with the kids.

Cloud computing means lots of different things, and much of it is hype. At Yieldex, we’ve been using cloud computing, specifically Amazon Web Services, as a key part of our infrastructure for the better part of a year, and we thought we’d pass on a few of our lessons learned. As you might expect, the services we use have trade-offs. If your challenge fits within the parameters, cloud computing can be a huge win, but it’s not the answer for everything.

All of these lessons are the result of the hard work of our entire engineering team, most notably Craig and Calvin. These guys are among the best in the world at scaling to solve enormous data and computation problems with a cloud infrastructure. We could not have built this company and these solutions without them.

For a startup, there are a number of compelling reasons to use a cloud infrastructure for virtually every new project. You don’t get locked into a long-term investment in hardware and data centers, it’s easy to experiment, and easy to change your mind and try a different approach. You don’t have to spend precious capital on servers and storage, wait days or weeks for them to arrive, and then spend a day or two setting them up. If your application scales horizontally, then you can scale additional customers, storage, and processing with minimal cost and time delay. All these things are touted by cloud providers, and basically boil down to: focus on your business, not your infrastructure.

Sometimes, however, you do need to focus on the infrastructure. We provide our customers with analytics and optimization based on our unique and proprietary DynamicIQ engine. Our first customer was a decent sized web property, and we were able to complete our DynamicIQ daily processing on several gigabytes of data using just one instance in less than an hour. Our next customer, however, was 10x the size. And the one after that, 10x more – hundreds of gigabytes per day. Fortunately, we had designed our DynamicIQ engine to easily parallelize across multiple instances. We spent some time learning how to start up instances, distribute jobs to them, and shut them back down again, but because we had designed the engine for this eventuality, we were able to use the cloud to cost-effectively scale to even the largest sites on the web.

We also have BusinessIQ, which is basically an application server that provides query processing and a user interface into our analytics. Initially we started with this server in the cloud too, but as we bumped up against other scalability issues, we found that the cloud doesn’t solve every problem. For example, we provide a sophisticated scenario analysis capability. To calculate a “what-if” scenario requires processing a huge amount of data in a very short time. For our larger customers, a single cloud instance did not have enough memory to perform this operation. Trying to stay true to the cloud paradigm, we implemented a distributed cache across multiple instances, but this didn’t work well because of limitations on I/O. We ended up having to go to a hybrid model, where we bought and hosted our own servers with large memory footprints, so we could provide this functionality.

We have been very happy users of the Amazon Web Services cloud, and not just because we won the award. We would not have been able to get our business of the ground with out the cost effective scalability of the Amazon infrastructure. While it’s not for every application, for the right application, it truly changes the game.

We are delighted to announce our Series B financing. The $8.5m round was led by Madrona Venture Group, a really smart group of investors. We met them through the Amazon AWS Start-up Challenge, and as it turns out they had been looking for a company like ours, so they had done a lot of research on the space. We were impressed by their industry knowledge and their experience, and are excited to be working with them.

We also are excited to have Amazon participate. They are the clear leader in their space, and their vision has proven out time and time again. We certainly hope they are right about this investment, too! Their AWS platform is what enables us to scale so cost-effectively.

Finally, a big thank-you to our Series A investors Sequel Venture Partners and First Round Capital, for their advice, counsel, introductions, and yes, their support in the Series B. We could not have build this complex and innovative technology without their support.

This is just the beginning for us. We are now well positioned to succeed, and our success is largely within our own control – just the way we like it. Great athletes always want the ball when the game is tight – we now have the ball, and we will win this game.

This is an exciting week for us at Yieldex, we are finally launching our first product, BusinessIQ! We’ve been working hard for quite a while on this, so it’s very liberating to be able to tell the world about it. Read our press release, or see the MediaPost article, for the details.

Having built enterprise software before, we know how important it is to have real customers banging on the product to make it solid. We are delighted to have Martha Stewart Living Omnimedia as our debut customer, and look forward to announcing more in due course.

MSLO has been a great beta partner for us, give us tons of constructive feedback and being patient through our inevitable growing pains. We have been able to iterate the product very rapidly to address their needs, and we are continuing to improve by leaps and bounds. We are excited by the value we are providing to them; we love seeing our hard work start to bear fruit.

Congrats to the team for this milestone – let’s enjoy it! Okay, that’s enough, get back to work.

The doom-and-gloom set have been getting a lot of press lately, and the conventional wisdom seems to be that display advertising will die off in favor of performance-based marketing. But don’t write the obituary yet. There’s an interesting new study from MarketingSherpa and ad:tech that surveys 1200 marketers and concludes that most actually plan to increase display ad spending in 2009. From the report summary:

The greatest shift in budgets is for behaviorally targeted ads. About one out of five marketers (21%) are cutting their budget while more than half (52%) are investing more money. Slightly more than 30% of the respondents said that behaviorally targeted ads were providing a great ROI, as noted in the first chart.

Surprisingly, 46% of marketers reported that they are increasing spending on rich media ads, despite the fact that more marketers reported that they deliver a poor ROI (27%) than a great ROI (23%), as noted in the second chart.

Traditional online ads will get more spending from 29% of marketers. That tops the 24% who said they’re cutting budgets in this area. This is surprising as well since about 1 out of 3 respondents (34%) said banner ads deliver a poor ROI and only 13% said they were great.

Perhaps the branding effect of the ads, while not directly attributable to revenue, is seen as vital. Almost half of marketers (47%) are holding steady in this category.

Surprisingly, this seems to suggest that marketers are not exactly running away from traditional online ads, but instead could actually be increasing their spend, particularly for rich media. And behaviorally targeted ads are typically display ads too, just targeted at audiences instead of content. So while performance-based advertising is very important to every marketer’s mix, let’s not lose sight of the fact display advertising, in various forms, is critical too.

The Richter Scales got a great response from the audience at the Crunchies, after making fun of just about every aspect of running a Web 2.0 company in these trying times. My contribution consisted mostly of showing up and not flubbing the couple lines they gave me – it’s great to be in a group with such talented people!