Sabermetrics & Search Bid Management; ‘Moneyballing’ the Market

It took some time for the baseball world to accept the forward thinking Bill James. His work in advanced statistics, now known as sabermetrics, attempts to quantify all aspects of baseball to determine why teams win and lose. James’ approach has been highly contested since the late 70’s, but is now becoming widely accepted as a stat revolution in baseball.

My curiosity with Search Engine Marketing (SEM) grew with the acceptance and advancement of sabermetrics in Major League Baseball in the early 2000’s. The similarities between SEM and baseball seemed quite natural to me and easy to correlate. You have a batter and a pitcher, similarly representative of your ad and a consumer. At-bats are impressions, hits are clicks, home runs are conversions and so forth. Slightly more quantifiable stats in baseball, such as Batting Average, Slugging Percentage and On Base Percentage equate very similarly to CTR, Revenue-Per-Keyword and Average Order Value for any given keyword.

The age of “big data” for marketers had me convinced there are new sabremetrics yet to be discovered or written for search which would provide a tactical advantage for my clients in the search auction. The logical correlation seemed obvious and as with baseball, sabremetrics could help smaller advertisers/ ball clubs compete against the large advertisers/ big market clubs as many of us enjoyed in “Moneyball.”

So I began hurling ideas at the whiteboard in the spirit of Old Hoss Radbourn and his 60 win season (minus the alcoholism and shooting out his own eye) – positives, negatives, extrapolating new stats, correlating stats by macro and micro trends and the findings were quite astonishing… as well as completely inconclusive.

Although it seems SEM is poised to enter the age of advertising sabermetrics, it quickly became more complicated than it was at first glance. The issue lies at the root of aggregating data, as with individual players. An individual player will average between 450-600 at-bats per season based upon games played and order in the lineup; however there is no change in competition besides the pitcher. Keywords, on the other hand, may have millions of impressions and in very different market conditions and at different ranks over a short period of time. Factors such as day parting, budget management, geo-targeting, seasonality, competition and current bidding algorithms make the environment too variable to quantify additional metrics beyond our current capacity. Although it would be easy to assume in the age of big data someone could ‘Moneyball’ the SEM space, it doesn’t appear likely. Actually, it appears nearly impossible.

By this point you are probably asking, “So what is the right approach and aren’t you going to share something insightful?” The answer almost appears too simple, considering the complexity of the issue, but it is as basic as implementing strategic, statistical, and active management.

Bid management needs an active eye – a daily glance ensuring you adapt to ongoing market changes. This means multiple rounds of bid optimizations at different points during the week. The statistical layer includes identifying a KPI (Key Performance Indicator) and leveraging the keywords driving your KPIs. The final, strategic layer comprises a variety of external market conditions such as seasonality, inventory and actualized return around various KPI’s to ensure you are maximizing actions for your given budget.

Sabermetrics may have changed the way GMs manage baseball teams and is even impacting other industries, but ad tech won’t be joining the party any time soon. Instead, we should embrace the fact that our industry only rewards those who actively participate in bid management, and those who succeed will earn their way to the top.

IgnitionOne’s Solution to Bid Management

Although developing new sabermetrics for search may not be a feasible solution, there are other technological approaches to addressing this challenge. IgnitionOne utilizes SPOT, which uses proprietary algorithms based on advanced statistical modeling technologies to accurately forecast keyword performance at varying spend levels. This can help marketers determine how aggressive bids should be based on goals and how to optimize spend by keyword to where it will receive maximum impact. By developing a technology solution which aims to provide insights at a level beyond what any new ‘search sabermetrics’ could provide, IgnitionOne can optimize keywords to achieve maximum performance at unprecedented scale.

Leave a Reply

Your email address will not be published. Required fields are marked *