Empirical Brewery Brews With (Data) Science

March 09, 2017

By Rob Arthur, March 09, 2017

Brewing has always been both art and science. The process of making beer is fundamentally chemistry, and big breweries maintain laboratories where they monitor the level of various compounds with incredible precision.

But the taste of a brew – how much and who likes it – is much more difficult to measure. Chicago’s Empirical Brewery is working to change that, applying the scientific method to quantify and improve the taste of their beer.

Empirical CEO Bill Hurley, a former Wall Street investment analyst, created the idea for Empirical in 2011. “At the time, there were probably 1200 breweries in the country… I wasn’t going to start one up unless I had a concept. The concept I came up with was experimentation, constantly incorporating the lessons we learned back into our beer.”

But it took a while to build the necessary infrastructure. In order to create a variety of beers to test against each other, Empirical needed equipment that would allow it to "constantly be brewing small-batch beers.”

At the brewery, Hurley shows me a set of five jacketed, one-barrel fermenters, assembled with a mix of off-the-shelf and custom-built components, all controlled by a Hackintosh laptop. Empirical uses these five fermenters as its laboratory, tweaking and tuning recipes before turning them into production beers on a larger 30-barrel system.

Steph Byce / Good Beer HuntingThe baby testing system controlled by computers is as beautiful as any.

It’s not as simple as brewing five kinds of beer and picking the best, though. To make their science rigorous, Empirical performs double-blind taste tests on the customers at the brewery, asking them to fill out a questionnaire in order to harvest their feedback. Neither the customer nor the bartender knows what’s being poured, and while the bartender can sometimes guess, they are under strict orders not to provide any information.

The data from the questionnaires is then analyzed and used to tune the recipes. Empirical’s new flagship, the Resonance cream ale, is an example of their process in action. They tried four different recipes on the public, before finding that there were two that clearly stood above the others. ”So that’s Mark Five sitting out there right now, a combination of two recipes that did exceptionally well in our experiment,” Hurley said.

Experimentation can help refine a beer, but the brewers need both a starting point and an end goal in mind. “The first time we brew a style, we try to stay right in the middle of the style,” Hurley said. From there, they iteratively adjust the beer’s recipe towards an interesting, unusual objective. “Any good baseball pitcher, they don’t pitch right down the middle… we try to stay away from the center of a normal style. That’s where we have fun.”

Empirical’s most popular beer, the Infinity IPA, is an example of the evolution a beer can go through. Beginning with a more standard IPA, Empirical gradually added hops to make it more and more flavorful. Hurley claims the current iteration uses “twice as much hops as the next beer in its category.”

But to balance that extreme hoppiness, Empirical produced a succession of recipes that improved on its drinkability by varying the yeasts and mix of hops. “We had to make it exceptionally dry, use a certain hop profile… You should be able to drink it again and again – you’ll be hammered – but as much alcohol as you can handle, and not have your taste buds overwhelmed.”

Steph Byce / Good Beer HuntingYour data makes the whole thing hum.

At times, Empirical’s data can produce truly surprising results. In one case, a mistake at the brewery resulted in an experimental fermenter full of beer with an excess of diacetyl. Diacetyl is a common compound in brewing and food which adds a buttery taste.

“Normally we’d throw it out, toss the whole batch… But instead, we decided to throw it into the experiment, see what happens.” Surprisingly, the diacetyl-laced IPA got “a massive positive response.” Unfortunately, diacetyl degrades naturally into another compound with the flavor of nail polish remover, so Empirical couldn’t reliably recreate the beer for commercial use.

There are also correlations between the characteristics of the drinker and the kinds of beer they like. In one questionnaire, they asked the customers to describe their diet. They found that people who ate more healthy diets tended to prefer beers with fruit adjuncts. In another case, they saw that gender was a major determinant of whether a customer would prefer a certain kind of beer – men preferred the more alcoholic beer.

And one variable that’s always important are the other styles of beer a particular drinker prefers. As you can see in the table below, those styles also vary greatly depending on age. Millenials seemingly prefer ciders and IPAs, while the older generations prefer lighter lagers. 

Mike Duesenberg / Good Beer Hunting. Data: Rob Arthur / IRIPreferences by generations reveal taste profiles that could fuel Empirical's choices.

Considerations like gender, diet and style preferences make the data analysis more complicated than simply using the highest-rated recipe. Given the diversity of tastes among drinkers, the beer with the best overall average may be the one that everyone hates the least, rather than the beer that they all love the most.

Instead of the average, Empirical tries to create beers that appeal very strongly to a subset of their audience. In one test, they compared their Infinity IPA to three other competitors’ beers in the Chicago market. All were highly rated, with one in particular universally-beloved, award winning, and retailing for about $14 a six pack. By average ranking, the Infinity IPA ranked third. “If you just totaled it up, you’d say ‘oh this beer came in third place, it’s a terrible beer,’” Hurley explained, “but the people that liked it, they really loved it.” The beer in fourth was the acclaimed $14 six pack, and it had the second-most top rankings. Infinity had the most top rankings. 

So far, Empirical’s beers have yet to garner the overwhelming praise of some Chicago-area breweries. But Hurley is confident that their scientific approach will lead to success. As they gather multiple years of data and better refine their methodology, Empirical will be able to more rapidly evolve and improve their beers.

Despite their novel approach, they have the same objective as many other brewers: “The goal is to be the best brewery in the country,” Hurley said.

ZX Ventures, a division within AB InBev, is an investor in October
Related Articles

This Oregon Brewery Is Betting on Braggot

Viking Braggot Co. is the only brewery in the country to exclusively brew braggots—a drink pioneered by the Vikings that combines hops and honey.