Jonathan Mendez is one of the most respected commentators on online advertising, and has extensive experience of working with some of biggest digital media companies. His blog optimize and prophesize is a great source for innovative thinking and is a must read for anyone working in the space. Mendez’s latest offering, Yieldbot, is a supply-side analytics company, which helps publishers garner better insights into user behaviour in order to boost performance from site advertising. Mendez recently spoke to ExchangeWire about the problems Yieldbot is looking to address, the dysfunctional state of the display eco system, and why the cookie gets so much unnecessary praise.
Can you provide an overview of what the Yieldbot is offering?
JM: Right now we are offering a turnkey performance channel for publishers using their data and monetizing their audience for their financial benefit.
How will Yieldbot help publishers?
JM: The most important thing we’re doing is shaping publisher data in a way that is actionable and valuable as a means to deliver relevance. This is data they are already collecting in their site analytics. We use the data to provide a channel that did not exist before. Most of our impressions (and revenues) are incremental.
Eventually our data shaping will unlock a world of potential value pubs have been unable to realize prior. Data is a publishers most valuable asset and at the end of the day everything we’re doing with it is about generating incrementally more revenue.
How would Yieldbot differentiate itself from the likes of supply-side optimisers like Admeld, Rubicon and Improve Digital?
JM: We fundamentally have nothing to do with network traffic or optimizing in the current display eco-sytem. We don’t care about CPM and inventory optimization. We’re not about prices and pages. We are about people and performance. Our goal is relevance. We are more an analytics technology focused on matching and pattern recognition than anything else.
Conceptually we take the reverse approach to targeting display than is taken now. We approach display the same way an optimizer would approach Search and that is start the ad matching with the visitor and their goals. A major problem with relevance in display is that advertisers start with an ad or offer and go fishing to find an audience that might bite. We go find the interests and intent first and then go find the ads and offers that match it. As Google has proven, it is much easier to deliver relevant matching that way and it can be scalable.
Where would you say Yieldbot sits in the present display eco-system?
JM: Nowhere, I hope. The current display eco-system was not built to deliver relevance. It was built to place magazine ads on the web. True relevance and thus true media valuation is stymied by the existing eco-system. We’re working on creating a new platform that is woven into the underlying data, content and actions of site visitor and matches ads against the intent and interests we harvest from that.
Can you explain how a publisher would integrate Yieldbot with their present ad server?
Ad servers are typically dumb. What we do is send the ad server a rule of what to serve once a visitor action or event on the publisher site qualifies them into one of our segments. So in effect, our segmentation rules make the ad server smart.
A recent post by AdAge tracked the ad dollar as it passed through the entire display eco-system. It estimated that out of $5 dollars of advertising media spend a publisher only earns $1. Do you think there are too many layers between the publisher and advertiser – and are these vendors offering real value?
JM: Unquestionably there are too many layers. I don’t think anyone would argue that. Jerry Neumann has the best piece on this I’ve read http://reactionwheel.blogspot.com/2010/02/duck-duck-goose-on-demand-side.html . On the flip side when you have so many layers there are vendors whose offering is all about value in the seams. Often the value is being delivered as a band-aid on a broken system. The interesting question as we progress in this environment is who is delivering value at both the beginning (advertiser) and end (publisher) of the stack.
The other thing about the display eco-system is that I don’t think most people in it understand what is coming from Google. There is real chance they will swallow up the entire buy side channel much the way they did in search. Having experienced what happened in search up close I see a ton of parallels.
You are looking for Yieldbot to create a new channel for publishers that’s based on performance. Can you explain what data the platform will use to ensure better ROI for advertiser and higher returns for the publisher?
JM: There is no better data to understand publisher visitors, interests and intentions than their site analytics. I’ve been looking at site analytics everyday for the last 12 years and the problem with them has never changed. It is just numbers. Numbers can be interesting but numbers are not actionable. One of the reasons I loved working on Offermatica (now Omniture Test& Target) was that every time you looked at the data it told you what to do. That was a giant step forward in the evolution of analytics. With Yieldbot we have taken another step forward. We not only tell you what to do, we do it for you -- and optimize based on the results.
Can you explain how your platform builds the segments that are being used to match ads against?
JM: Yieldbot allows a marketer to create segments using keywords around interest & intent based on the referrer data, site data, page-level and keyword level semantic analysis. We use a variable look back period in the data to estimate the size of the segments and once segments are created we use trending analysis to optimize and send alerts about possible new segments or sub-segments that may emerge. Anyone proficient with AdWords would feel very much at home driving the segment creation but instead of an index of the web as whole we create an index for each publisher based on what we call the “user taxonomy” of the site.
Has your extensive experience in landing page optimisation informed the development of the Yieldbot offering?
JM: Absolutely. The origins of yieldbot go back 2 years when I had the idea to try and do the same things I was doing with dynamic landing page optimization in a 300x250 display ad. I was successful doing that but the problem was that it wasn’t scalable. Everything became a customer solution – just like building dynamic landing pages are. Still, the principles of dynamic segmentation and realtime targeting that make dynamic landing pages so successful are completely baked into Yieldbot as are auto-optimization & A/B testing of the creative.
You recently wrote a post about “the link” and the cookie, arguing that “the link” is often overlooked as a valuable web attribute by marketers. What do you see as the cookie’s major drawbacks?
JM: In all my years of targeting I’ve found that knowing what someone has done matters a little. Knowing what someone is doing matters a lot. So for cookies the major drawback from a performance standpoint is the temporal degradation of the interest & intent. Google uses a 2-hour lifespan for their retargeting cookies. I heard the other day that another well-known re-targeter uses a 15-minute window. What we’ve known since the Aristotle is the most important part of persuasion is timing. The Greeks word for it was Kairos. Of course cookies are under attack by privacy advocates as you are well aware of in Europe. We wanted to build a system that would never have privacy issues for publishers, brands or their audience.
What type of publishers will Yieldbot be aimed at the initial stages of the launch.
JM: Minimally a few hundred thousands unique visitors a month but more than the size of the site it’s about the value and amount of segments we can create for publishers that will determine how successful they can be. But of course volume helps.