Auctioneers … it’s A/B testing, not A testing

Tags

, , , , , , , , , , , , ,

A/B testing is a useful tool to compare two different ways of doing otherwise the same thing … to see which is better/worse. For example, I could A/B test eating an ice cream cone. For “A” I eat it top down, and “B” I eat it bottom up. Then I evaluate and decide which works better: A or B.

The widely held definition of A/B Testing is:

A/B testing is a controlled experiment with two variants, A and B. It is a form of statistical hypothesis testing or “two-sample hypothesis testing” as used in the field of statistics.

So for any auctioneer to say, “See, this sold for [whatever] with a buyer’s premium so buyer’s premiums work well” ignores the question, “What would that same asset have sold for without a buyer’s premium?” And, “See, this sold for [whatever] “with reserve” so with reserve auctions work well” ignores the question, “What would it have sold for if it had been put up “without reserve?”

Ryan George has talked about A/B testing numerous times including a blog here: http://www.ryangeorge.net/backupbidders/ Not once have I found Ryan advocating “A” testing … nor virtually anyone else for that matter.

Here’s the important lesson: If what you just did (absolute simulcast auction with buyer’s premium) worked well, you should ask yourself if it would have worked better with reserve, live-only, online-only, or with no buyer’s premium. Further, unless you have (A/B) tested all those scenarios, you don’t have those answers.

Fortunately, we as auctioneers don’t have to necessarily do all this testing because other auctioneers have done it for us. We can leverage the experiences of veteran auctioneers in classes held at state and national auctioneer events, seminars and the like. For example, when Ryan George says, “This type of Facebook marketing works better than this other option,” you can probably consider that a fact, as Ryan has (A/B) tested it.

I was discussing in a class the other day the Supreme Court of the United States decision in 1926 that if a buyer at auction is held to have purchased something “as-is” then he would have to have had the reasonable opportunity to preview that asset before that purchase. A student countered that he sells “as-is” all the time but doesn’t allow any previews. While this isn’t strictly A/B testing, if A is allowing a preview and B is not, I can assure all auctioneers A will likely work better.

Lastly, I remain on the lookout for a single “without reserve” auction being advertised as “with reserve” in order to increase participation — and/or an auction without a buyer’s premium being advertised as “buyer’s premium charged” in order to increase participation. Even without a strict A/B test, isn’t it easy to recognize how auctioneers around the country know these differences effect participation … and thus price? Far more empirical evidence than the aforementioned anecdotal evidence for sure.

Mike Brandly, Auctioneer, CAI, CAS, AARE has been an auctioneer and certified appraiser for over 30 years. His company’s auctions are located at: Mike Brandly, Auctioneer, RES Auction Services and Goodwill Columbus Car Auction. He serves as Distinguished Faculty at Hondros College, Executive Director of The Ohio Auction School, an Instructor at the National Auctioneers Association’s Designation Academy and America’s Auction Academy. He is faculty at the Certified Auctioneers Institute held at Indiana University and is approved by the The Supreme Court of Ohio for attorney education.