Categories - Featured
1. Tell us a little about your background and how you ended up in search marketing.
I have a Ph.D in engineering from the University of Massachusetts at Amherst in the field of computational statistical thermodynamics (CST). It’s a mouthful, but essentially it means using computers (‘computational’) and math (‘statistical’) to study problems related to phase transitions, chemical physics, power-law behaviors, rare-event statistics, and so on.
I found The Search Agency through a Google Maps search. Most people go to Maps and type in things like “sushi restaurant” or “movie theater”. I type in things like “stochastic optimization” and “Shannon entropy”. A result popped up for a Craiglist listing for The Search Agency, which hires a lot of people with strong technical backgrounds.
Search engine marketing was (and still is) a rapidly growing field and also rapidly growing in technological sophistication. I looked into PageRank and learned that it’s just an iterated, matrix-based alternative to a Markov simulation. I looked into the work by Barabasi and Bianconi on modeling link formation in a growing web of homepages and they openly acknowledge taking equations, and even notation, from CST.
Then, I looked into the systems used by search marketing firms and they were all dirt-simple or rather naive: rules bidding, position bidding, modern portfolio theory. I thought, there’s a real contribution that physics can make to online marketing. When you look at the companies that really “get it”, The Search Agency (TSA), Rimm-Kaufmann Group, Marin Software and Rocket Fuel, they are all full of people with engineering and science backgrounds.
2. What is your role at The Search Agency? What new projects are you most excited about?
My job title is ‘Manager of AdMax Research and Development’, where AdMax is TSA’s proprietary online marketing platform which handles paid search, SEO, display advertising and so on across publishers.
I’m excited about TSA’s project to re-vamp AdMax’s conversion attribution functionality. I obviously can’t reveal much about it, but I think that much of the industry has been thinking about the problem of properly attributing conversion events throughout the entire conversion path in a way that’s fundamentally orthogonal to a proper attribution methodology.
I’m intrigued by Google’s Bid Simulator (BS) feature and the wealth of information it is providing on identifying non-equilibrium bid landscapes, transitory dynamics and profit-maximizing vs. relative benefit-maximizing bidding. Frankly I don’t understand why there isn’t orders of magnitude more interest in the industry in this tool.
I’ve been working with some graduate students at Brown University on simulating model paid search environments and identifying optimal account management strategies as part of the Trading Agent Competition I’ve blogged about before.
Finally, as I just said, I’m always excited about sharing my perspective on search marketing through my writings at The Search Agents’ blog, Search Engine Land and Search Engine Journal. TSA is very supportive of any ‘over and above the call of duty’ activities that employees engage in and I’m very lucky to be able to devote as much effort to it as I have been.
3. The Search Agency just published a white paper on the various approaches to bid optimization. What are some of the key takeaways for marketers trying to learn more about this area of paid search?
The single biggest lesson, I think, is that even though many firms sell themselves as the single right solution to bid management, there no “Holy Grail” in paid search. Depending on the situation, there are better ways and worse ways. For firms with large campaigns and high spending levels, people with spreadsheets simply can’t hand the volume and complexity of the data. But, smaller, lower-volume accounts face uncertainty due to fundamental limits in the ability to determine conversion rates, value-per-conversion and so forth.
At TSA, we tend to be what I call ‘algorithmic agnostics’ – we use the most appropriate solution for the given problem. That leads to what I call the ‘hybrid approach’ to bidding. It incorporates a statistics-based approach founded in physics and network theory when that is most appropriate, steps down to a rules-based or portfolio theory-based approach when those are most appropriate, and finally to a position-based or fixed-bidding approach when those are best.
In the end, though, you can’t outfox math. I always find it comical when those without any technology make irrational anti-technology arguments. So leave the math up to the statistical algorithms and let the people add value in areas where the technology has not yet surpassed human brainpower and judgment.
4. What are some of your interests outside of the office?
I help to do online marketing for OMGbabycards, a company that sells gorgeous, traditional-style birth announcement cards.
A couple of years ago I bought an old farmhouse and have been slowly renovating it since. I’m pasture-raising rabbits and helping a neighbor (and fellow dirtworker) with his organic, grass-fed beef business. So, I spend a lot of time doing typical small farm things: woodworking, composting, tweaking multidimensional simulation routines, caring for animals.
And the past few months I’ve been publishing my eponymous personal blog, where I write about things that are not quite right for The Search Agents, like using Gompertz’s equation to model the growth rate of my chickens and things like that.