Keg 𝑣 Bottle & Can — a natural experiment at the AIBAs

Histogram of data on the relative performance of beers entered into the 2024 AIBAs in keg and pack format. Wide green bars with hand-written numbering show a mostly "normal" distribution, with a big spike in the middle, curving out to either side, with a slight bias to the bars on the right. (The data is in a table, below.)
You can tell it’s a nerdy one when it starts with a diagram rather than a photo

I’m going to try and convince you that this an interesting graph. It charts the relative performance, at last year’s Australian International Beer Awards, of all the beers that were entered twice; once in keg, and again packaged. This year’s judging is currently underway in Melbourne, so the process is on my mind, and this is a rare opportunity to test a common belief that breweries more able to sacrifice a whole keg are at an advantage.1 Short answer: only a little, if at all.

Most competitions don’t give out full data on all their entries; it’s more common to just publish the winners without publicly listing the ‘losers’ — and I’ve had a lot of fun with the fact we get more information out of the big Australasian events. But in New Zealand, entering a beer twice is forbidden;2 in Australia, it’s pretty common: about 20% of all the local entries were present in two formats,3 and 64 breweries, of all sizes, took that option with an average of 4.5 beers doubled-up each.

In terms of my usual post-awards analysis, kegged beer did indeed fare slightly but noticeably better overall at the 2024 AIBAs — and a tiny majority (11 out of 21) of the style-champion trophies (e.g., Best Pilsner) went to beers that had been judged off tap, a little more than ‘expected’ from the fact 40% of all entries came in kegs. (And please take a moment to contemplate the logistics of nine hundred of those arriving.)

AllKegCanBott.
Medal percentage74.779.872.267.7
Points per entry1.221.361.131.15

But I was curious to see how those twice-entered beers fared against themselves. So I singled them out and worked out the effective Keg Advantage — graphed at the top of this post, and tabled below.4 In the middle are all the beers where their results were equal both times (however good or bad), while +1 means the kegged version outperformed its packaged sibling by one medal class (a silver over a bronze, say), and the sole -3 was a bottled beer that won gold but didn’t medal at all in keg.

-3-2-1=+1+2+3
186510685245

So “no difference” is the most-common result and 87% of the pairs landed within one rank of each other — but every possible combination of results did actually happen. The slightly lopsided graph and somewhat bigger numbers on the right shows some overall advantage. The average of the whole set was about 0.22: all else being equal, entering a beer in keg got it just over a fifth of the way towards the next-best rank than it would’ve achieved otherwise.

But all else is rarely equal, and there are a bunch of caveats to consider. First, I’ve only got medal data. They don’t release the actual scores (out of 20 possible points),5 so some of those “equal” results could be a little way apart and some of those +/-1 pairs might only actually differ by a tiny fraction. Second, there’s no guarantee that the keg and bottle stock put forward for judging come from the same batch. It seems the most likely scenario — partially because of the appeal of getting independent feedback on packing processes — but breweries can and do mix and match, maybe after in-house testing, to increase their chances of a headline win.

It was Luke that prompted me to finally go digging in the AIBAs data, and I feel like it was his book (with its apt title) lurking on my shelves that kept nagging at me until I did this

The other variable besides packaging is the judging itself. Some semi-random personal variation is inevitable, even though panels of judges are used and their scores combined to smooth off outliers. The sequencing of beers through the day or whether one follows a particularly good or bad other sample, could easily have enough of an effect to knock something up or down a rung. Palate fatigue is real, even if the ‘hungry judge’ phenomenon isn’t. And the brain would inevitably bristle against too long a run of similar scores, even if that’s what the beers presented to it ‘objectively’ deserved. Wine competitions have lately been subjected to somewhat-controversial examination; it’d be interesting to see something similar done with beer — if you have any examples, let me know. Maybe the AIBAs or NZBAs would be a good place to start. If I was them, I’d already be running experiments behind the scenes…

As it is, there’s just not enough data and too many confounds to go really wild with the conclusions, here. This is your classic “in summary, more research is needed” kind of ending, I’m afraid. But it looks interesting enough to pursue and repeat with next week’s 2025 results.

Initially, it looks like Keg Advantage is real, but not as massive as people might think. So that’s worth pushing back against, because it does seem to discourage some would-be entrants. And perhaps competitions (like the NZBAs) that worry about flagging entry numbers should consider removing the rules against entering beers in multiple formats. (Especially if that’s paired with more sophisticated ways to crown your “champions” that avoid rewarding entry-spamming.)

Meanwhile, it’s great that this much data is available from the AIBAs in the first place. More competitions should follow their lead. The World Beer Cup only tells us how many non-medalling beers were in each category, and the World Beer Awards don’t even give you that.6 With transparency, these awards can be a kind of peer review; without it, they’re just PR.


  1. In a footnote, because I’m not sure who needs to hear this but might also be interested in a post this niche: Packaging (whether in cans or bottles) introduces more risk to the beer: oxygen or microbes getting in during the process, and the smaller volume makes changes in storing temperature more problematic. Plus, in the case of bottles, the danger from light — though dark glass mostly takes care of that. So kegged beer faces fewer potential sources of demerits, but it’s expensive to give up that much stock, so bigger or richer or more-determined companies have a head start.
  2. “A beer may be only entered once in any form (i.e. only keg or bottle, not both), once in name only (i.e. not the same beer brewed at different brewery / locations) and only entered in one class.” — 2024 NZBA Entry Guide, p15
  3. Only five overseas beers, though — despite a total of 78 kegged entries coming in from international entrants, so it’s not just the logistics that stopped them. Perhaps it’s partly that most competitions don’t allow it, so breweries didn’t really consider it.
  4. The magnificent TablePress plugin didn’t have to work nearly so hard as usual, here, but it remains magnificent.
  5. Melbourne Royal 2024 AIBA Entry Booklet, p36
  6. When Cassels were crowing about being the New Zealand winner last year, it took a few emails before the organisers would give me the crucial context: they were only up against one other NZ entrant. More a cointoss than a contest.

Have at it: