I just wanted to take a moment on the blog here to really expand at length where I get the numbers and times I state with (some) confidence in discussing lifespan.
I've always been super curious about coffees' "freshness" lifespan, because it seems pretty subjective and seems a very contentious thing in rarified coffee circles.
I mean, we've all seen best-by dates claiming months on end, and we've all got that coffee-snob bud who wants you to believe all worthy coffee is 'dead' by six days old. They're probably not both right, at the very least.
To me, that hyper-shortened lifespan seems sold like The Emperors New ______; where if my beans or your beans or their beans are purported to last longer - "well, they couldn't have been that good anyway". But it's similarly clear that some folks are really overstating the lifespan of their products and are relying on an underinformed or apathetic consumer to 'get away with' their claims.
So I set out to test this thing out a little. I used my apartment kitchen, my mum's dinner parties, foot traffic at the plant and customers caught unawares ... essentially, I was forcing many little cups of coffee at anyone who'd come within a serving trays' reach of me.
With a self-reported survey of 'coffee acumen' - priming question was "Speaking broadly, how knowledgeable or dedicated would you say your relationship with coffee is?" most people came back with some variation of "not very knowledgeable, very dedicated", even from people I would have scored 'higher' on knowledge. However, I feel I do have enough total pool to have a decent representation of 'average consumers,' 'average specialty coffee consumers' and 'highly coffee-passionate consumers' or 'experts.'
I "ran" three groups, like I was doing Real Science or something. All three groups got the same range of samples: five cups of the same product at one-week lifespan intervals. The first group was told we were QA-ing separate, similar, components for an end blend, and they should report what they think of each for individual quality. The second group was told we were assessing potential purchases, and they should rank by preference in order, and note any 'jumps' or 'drops' in quality between samples. The third was told we were assessing lifespan, and they should tell us which samples they found staleness present in. This group was sampled a very fresh sample and a very (two months) stale sample to calibrate their palates.
The first two groups came out with remarkably parallel results. For the coffee I'd used on that round, most subjects from either group were scoring weeks 2-4 quite high and descending in order, then week 1 slightly lower, and week five as an 'very poor' outlier. I think a safe conclusion to pull here is that folks preferred weeks 2-4 relatively interchangeably and found notable decline in the fifth week.
The group primed to check samples for freshness or 'look for' staleness found staleness much earlier. They generally reported decline from the three-week old sample onward, with some outlier participants even scoring some staleness in the two-week-old sample. I think a safe conclusion to pull from this group is that some staleness can be detected at three weeks from most consumers.
But what gets most interesting is when we combine what our three groups have told us and see how that works loose. If you're actively looking for staleness, you're more likely to find it earlier in 'younger' samples than when you're not looking for it. This could be a reasonable explanation why some companies swear by seven days while others will quote thirty-five - a difference in approach, values, and sampling methods can return very different results. Weeks three and four scored quite highly (above the one-week) sample when subjects were assessing quality, but staleness was consistently reported at those same ages from the staleness-primed group. This could be a reasonable hint that staling starts fairly early but being perceptible while subjects are looking for it doesn't mean that same 'amount' of staleness negatively impacts overall enjoyment until much later.
I'd like to note that I worked very hard to control for levels of coffee knowledge and aptitude across groups. Highly coffee literate subjects, 'experts' if you will, were not more or less susceptible to the priming effects described in the difference between groups 1 & 2 and group 3. They generally had fairly similar quality or preferential rankings to more 'lay' consumers, while those participating in group 3 would note staleness slightly earlier - but I don't have enough data to say for sure that one's not stat noise.
Running follow-up tests to see how groups play out across different kinds of coffee and roast profiles is a much bigger task, and one that's sort of a long-term project. So far it looks like different beans have different trends, all within the expected "it starts tasting boring some time between three weeks and five, by five its almost always done" with no huge trends across roast level and/or origin.
In both cases, I do not have a 'clean' enough experimental setup or a large enough body of data to draw any truely academic-grade conclusions. I might have enough to feel comfortable speculating and suggesting, but that's my own interpretations and results so far are not solid enough data to escape the potential influence of opinion.
Because the third group needed to be 'calibrated' with a earlier service (two additional cups served prior to the core five) it was more challenging to fit folks into our schedule, or tests into theirs - I did end up with a sample group about 2/3s the size of the other two.