One of the most interesting developments in craft business models in recent years has been the increasing importance of on-site beer sales, particularly for small production breweries. On-site sales at the brewery are not new to craft. Until the last few years, brewpubs routinely represented 10 percent or more of craft’s total production and the vast majority of brewpub sales have always occurred on-site. What is new is the increasing importance of on-site sales for non-restaurant breweries.

In 2014, I estimated that on-site sales (on our survey we ask for “volume removed tax-determined for use at tavern on brewery premises”) at around 1.5 million barrels. This year, that number appears to have increased slightly faster than the overall category, and I am estimating it at 1.75 million barrels (craft only, so that volume would be a touch higher if you were talking about the overall beer market).

That works out to roughly seven percent of craft’s domestic sales, but still less than one percent of overall beer sales. If we look at it versus the on-premise market, that’s still about five percent of overall on-premise volume, though it may be as much as 20 percent of craft’s total on-premise volume (I’m assuming it’s actually less than that since some breweries probably included packaged sales in their figure). Remember that this includes brewpubs, who represent about half the total on-site volume, so if you’re just talking about the packaging breweries, the figures listed above go down considerably.

Because this is the second year we’ve gathered this data, we also have the luxury of looking at a “same brewery” group–the 565 breweries that provided both production and on-site sales data both in 2014 and 2015.

Microbrewery On-Site Beer Sales

For microbreweries, the data shows what I’ve long argued, that micros tend to shift their sales mix into distribution as they grow, and so on-site sales are more important as they start up.

Micros, in this particular sub-sample, saw their on-site sales as a share of total production decline from 16.7 percent to 15.7 percent, as distribution sales grew faster than on-site sales (33.7 percent vs 24.9 percent).

Brewpub On-Site Beer Sales

What was perhaps more interesting in the data was that the brewpubs in this year-over-year sub-sample also saw more sales going into distribution. Brewpubs who provided data both years saw on-site sales go from 73.7 percent to 66.4 percent, as their sales were largely flat on-site (three percent) growth, but they saw strong growth outside the brewery (46.5 percent).

This certainly could be a biased subset (more on this in a second), but regardless, it shows that the lines between the microbrewery and brewpub business models continue to blur as more brewpubs enter production and distribution.

Digging Deeper into the Stats

Fair warning: the rest of this post is more statistical/methodological in nature…

Alright, if you’re still reading, how did I come up with these numbers and what are the potential flaws in this analysis?

The numbers are based on our annual BIPS survey–where 1,497 breweries told us what they sold at their brewery.

The biggest issue with this data is that it’s likely that there is some form of selection bias in the sample of breweries that provide their on-site sales data during our annual survey, in other words, the breweries that filled out this section are different on average from the overall population. These biases come from both the selection of breweries that chose to fill out the survey in the first place, as well as the breweries that filled out that portion of the survey (some percentage of breweries just send in their production numbers and leave the rest blank).

Would such a bias mean that the “real” number is higher or lower than what we found? It’s hard to know. First, smaller breweries, who are more likely to have a higher percent of on-site sales, are less likely to fill out the survey in the first place. That would suggest that the real number may be higher. But, sifting through the data, my hunch is that there is a selection bias from two additional factors, which push in different directions:

• Brewpubs that sell 100 percent on-site may be less likely to fill out the on-site portion
• Packaging breweries that sell some percentage on-site are more likely to fill out the on-site portion. Some breweries fill in zero, but not as many as I would expect, particularly in states where it’s illegal to have direct sales.

I’ve based my estimate of 1.75 million barrels on three factors, two of which come to similar conclusions.

1. Taking the percent of on-site sales by craft category (brewpub, micro, regional) and then multiplying it by the volume of each
2. Taking the “same brewery” growth percentage, and multiplying it by last year’s volume estimate (and then adjusting upward a bit for new breweries)

Both of these methods come up with very similar estimates of between 1.7-1.8 million barrels, hence the 1.75 million barrel estimate.

The third method is to dive into the California state data, which includes the volume sold at the brewery. You can then take the percentage of volume breweries sold on-site and multiply it by both craft volume and the overall industry. A sub-strategy here is to look at California “same breweries” and see their growth. We don’t have the data from annual filers in California yet, but this came up with similar same-store growth rate as seen in our sample, and a similar overall on-site growth rate as seen nationally.

Using the California numbers does come up with a lower total on-site volume estimate, but I think that illustrates an issue with how the data is collected–some of the volume sold on-site is being put in a different tax determined category. If you watch the TTB beer data, you may have seen the same thing–the “tax determined, premises use” seems suspiciously low.

I suspect that many brewers are reporting kegs they sell at their brewery as “tax determined, in kegs,” rather than “tax determined, premises use” (which they are reporting when it comes from brite tanks). I see this gap in the data breweries provide to the BA, and the data they provide to California. The total numbers line up, but the on-site sales are sometimes fairly different. Is that how your brewery does it? Let me know.

So the California numbers don’t confirm the volume figures but do confirm the overall growth story. Taproom sales are growing at 15-20 percent, slightly faster than the overall craft industry. If that pace keeps up, 2016 could be the year on-site beer sales reach one percent of overall beer sales.