What exactly are we to make of this experiment and these results? There are several obvious conclusions which I will list below, and endless speculation which I will leave to the readers of this article.
Two months is safe: Freshly roasted coffee that is immediately frozen after roasting in a near airtight container in a very cold freezer, can be kept undisturbed in the freezer for at least 2 months and be expected to produce espressos that are not obviously inferior to those made from fresh coffee that has never been frozen.
Freezing does not accelerate staling after defrosting: At least over a period of time extending to about 8 days after roasting, using the roasting and freezing procedure used here, there was no evidence that previously frozen coffee deteriorates more quickly after defrosting than does coffee that has never been frozen.
This study was primarily designed to evaluate freezing as a method of coffee preservation available to the average home consumer. As such, it has demonstrated that freezing, done shortly after roasting in a very cold freezer delays staling for at least two months and hence extends shelf life for at least that long.
Two aspects of this study could present difficulties for some readers and they deserve clarification. First, the coffee was frozen immediately after roasting, and those who are reliant on parcel delivery services or who cannot buy just-out-of-the-roaster coffee must compromise on that. The second is that not all freezers, especially freezer compartments of refrigerators, can maintain very cold temperatures. Self-defrosting freezers are especially problematic in that regard. Nonetheless, I chose to use a "best case scenario," that is, immediate freezing in a very cold freezer, to study the impact of freezing for coffee preservation. Had I chosen otherwise and had results been different, we would not have known whether the compromised results were the result of delay in freezing or inadequately cold freezer temperatures. The data in this study can be augmented by reports of readers who have used less rigorous approaches, and I solicit their findings in the comment thread following this article.
Jim Schulman has previously communicated his results of some informal coffee "cupping" experiments he has done with frozen coffee. Cupping is of course a far different process than is espresso making and some would say that one can have more precision in cupping than one can have in judging coffee served as espresso. Jim reported that although he could tell previously frozen from never frozen coffee more often than if by chance, the differences were subtle and defied characterization along the lines of anything "systematic" or "easily describable." He felt he was no better in discriminating between frozen and never frozen than he would be in detecting subtle differences between different roast batches of the same coffee, neither of which had ever been frozen. I should add that Jim does not have a freezer that holds very cold temperatures like mine does, however he did freeze his coffee immediately after he roasted it.
Jim's observations imply that the observations in this study can be extrapolated to include coffee that is frozen and used in preparations other than espresso. This would require further efforts to prove, should someone be sufficiently motivated to try to test it.
The presence of more than one espresso machine and more than one grinder introduced other potential variation into the study that was controlled for by proper "balancing" within the experimental design. Because the study was balanced in this way, it enables us to look at these other variables independently to learn other things from this experiment. Below are some conclusions reached as result of further data analysis that includes these factors.
Old versus new grinder burrs: There is no obvious, tastable benefit from changing grinder burrs after moderate wear as was done on one of the two grinders used in this experiment. Some have speculated that changing planar grinder burrs much more frequently than is the stated life of those burrs would produce a benefit in espresso. We did not detect any such difference and therefore believe that very frequent burr changes, much earlier than the stated lifespan of the burrs, has no detectable benefit if taste is what is being measured.
Rotary versus vibratory pumps: This study is the third independent experiment which compares the output of two nearly identical espresso machines that differ primarily by the type of pump within them, one with a vibratory pump and the other with a rotary pump. As in the other two studies,  and , no consistently tastable difference was detected. Therefore, all suppositions that rotary pumps produce superior shots to those made by vibratory pumps must be regarded as unproven barring further experimentation.
Tasters do have measurable preferences: There was only one obvious interaction among any of these variables, which was that one taster, Randy, rated the coffee that had previously been frozen for 4 weeks with lower scores when he tasted it on his second session, 3 days after the first. In addition, there were several uninterpretable ones including such things as one taster liking one grinder/machine/coffee combination more than others. The effects are "obvious" in a numerical sense, which is why they are significant. However, they have no bearing on the overall result. As such they simply make the point that on some very specific combinations of machines, grinders, and coffees, some tasters will have idiosyncratic preferences. The whole point of using multiple tasters is to cancel such idiosyncrasies out. However, the presence of interaction effects disallow blanket statements such as "No taster will ever detect a difference regardless of machine, grinder and coffee being frozen." They prove there is a difference between fresh and frozen that comes out in rare and idiosyncratic circumstances. This observation in no way counteracts the fact that the overall results offer no evidence whatsoever that there is a systematic taste difference.
The wording of the last point is taken directly from Jim Schulman's commentary on the data analysis, for which I am grateful and without which this study could not have been done.
Finally, some reviewers of earlier versions of this manuscript raised the issue of extraction percentages and how this might have effected the results. It is true that the shots tasted in this study tended towards the ristretto, and the double baskets were "overdosed" with approximately 18 grams of ground coffee. Because this was done for all 128 shots tasted (64 shot pairs), in order to invalidate the conclusions one would need to come up with a hypothesis explaining how grinder fineness and dosing are interrelated with freezing. I cannot produce a feasible explanation of why this would be the case, and no matter how the shots were prepared some readers would feel it was not representative of THEIR own approach. The most important thing is that the preparation was held constant, so the impact of the technique I used should cancel itself out in the observed results.
Freezing is a viable method of preserving the freshness of very fresh coffee. Exactly how long the usable lifespan of coffee can be extended with freezing is unknown, although we do know that if frozen immediately that lifespan extends to at least 8 weeks. In this experiment, a very specific methodology was employed, and exactly how far one can deviate from it and expect to get good results is unclear. The previously frozen coffee we used was frozen immediately after roasting, within about an hour, in semi-airtight packaging in a very cold freezer (about -15°F / -26°C). It was then defrosted, only once, within the same packaging before it was exposed to outside air, reducing or eliminating the possibility of condensation.
If the coffee one contemplates freezing is not "fresh" to begin with, it is doubtful that freezing will do much of anything positive. To me this means that freezing is probably of no value when dealing with purchased coffee of uncertain age. In the case of coffee that has partially degassed, that is perhaps several days out of the roaster, it is unclear from this experiment whether freezing will extend shelf life significantly. Since this is going to be a fairly common scenario for home espresso enthusiasts ordering online, it deserves further comment.
For the homeroaster or person who can buy fine quality extremely fresh roasted coffee locally, it makes no sense to degass the coffee at all before freezing as this adds nothing and the coffee can be degassed later, after defrosting. On the other hand, if you are buying roasted coffee that has to be shipped to your door, then you will need to test for yourselves how well this works with the beans that you purchase. My opinion, which is not supported by any data (since I have none) is that freezing fresh coffee that is several days out of the roaster should extend shelf life by at least a few weeks. I would encourage readers who use this approach to give us the benefit of their own experiences in comments you make on the thread that follows this article. Similarly, if you don't have a very cold freezer, it stands to reason that the amount of time that freshness in coffee can be preserved will be less. How much less I do not know, but perhaps some readers will have their own observations that will prove useful.
If you are concerned about what sort of container you should use for freezing coffee, it obviously needs to be something that is relatively airtight and that can tolerate the conditions present in a freezer, and the temperature stress in going from room temperature to very cold and back again to room temperature. I generally use Mason type canning jars or recycled jars from grocery products that will close with a tight seal; I fill them up as full as possible to minimize the remaining air that is present. I have also used certain types of commercial plastic coffee bags that can be sealed and if valves are present I tape over them. If you purchase coffee that is already packaged in a sturdy valve bag you could simply tape over the valve and toss it directly into the freezer. I would however suggest that whatever container you choose, it be sized to allow you to consume all of the contents within a reasonable period, say 1 week, without having to open the bag and return some of the contents to the freezer; doing so risks condensation on the beans which could theoretically cause damage.