An experiment to determine freshness

Freshness is one of those difficult terms in coffee because it is often considered quite subjective.

However I was thinking about brewing stale coffee as espresso, and then thinking about measuring filter brewed coffee and an idea cropped up.

To me, when I brew stale coffee as espresso it seems that there are a lot less solubles in the brew – the pour looks very pale very quickly, if not all the way through the shot. So based on this observation I propose an experiment using filter coffee.

Roast up 5 kilos of a coffee and then set up to brew as filter coffee so that the percentage of extraction is around 20% for the chosen weight of coffee. (The brewing charts may differ on ideal strength but they all agree that 18-22% of solids extracted is considered ideal.) Let’s say we use a Nordic dose of 70g per litre.

Once this grind has been set for that dose then 3 brews a day are done for that recipe with a fixed grind (the grinder kept cool – no back to back grindings). The resulting brews are then measured for TDS, as well as the brew water so an accurate log of extraction percentages could be kept. This daily experiment is then repeated for 30 days and the results graphically logged.

The coffee could be kept a number of different ways for each type of experiment – craft packaged, valve packed, valve packed and nitrogen flushed.

I would hypothesize that though everything was kept constant (grind/dose/brew temp and volume) as the coffee aged it would start to be more difficult to extract. Whether you chose to see the point at which the coffee drops below 18% as the point it is stale, or perhaps another point that has some repeated statistical significance I don’t know. However I think that one could likely find enough data to give a reasonably accurate shelf life for truly fresh coffee.

It would also be interesting to couple the metering tests with cupping, but I worry that it would be too difficult to be objective/difficult to set up well.

I am sure this has flaws and would welcome any input on this before I start to play with it. (Plus I need a TDS meter!)

Thoughts in the comments please?

11 Comments An experiment to determine freshness

  1. AndyS

    Because you seek to avoid the difficulties in assessing “staleness,” (which is of course highly subjective), you’ve designed an experiment to objectively measure how solubles extraction varies as the coffee ages. Although solubles extraction and staleness are probably somewhat correlated, it seems like a huge stretch to make them equivalent.

    What if you found that the solubles extraction could be kept constant if one ground the coffee finer and finer as it aged? Eventually the coffee would still taste stale even though it measured the same TDS.

  2. James Hoffmann

    I don’t think that there will be an absolute correlation but it would be interesting to see how long the coffee “held” a consistent extraction (if at all) and how dramatic the drop-off would be (if at all).

    I agree that the test could be changed and results achieved through changing the grind, though I also wonder how dramatic the grind changes would have to be to compensate for the staling.

  3. Ian Clark

    If I understand correctly, it would become more difficult to reach a 20% target (or rather it would take more time to get there) as the coffee ages because the ‘desirable’ solubles present in fresh coffee are more readily soluble than the bitters/narlies that take up a greater percentage of the total solids in aged or ‘more stale’ coffee. Without these highly soluble particles, which have oxidized or otherwise left the building, it will naturally take longer to achieve a certain TDS.

    In these terms I can see how it would be possible to objectively define one particular coffee’s ‘stale’ time threshold under one particular form of packaging only. This is of course assuming that we can define 18% TDS in 4 minutes under blah blah blah conditions as objectively stale – this is still a subjective determination based on the expectations of the drinker and the amount of sugar the have in the cupboard.

    My concern is that the TDS wouldn’t actually change that dramatically. The aromatic portion of the chemicals in coffee are a very small percentage of the total. As far as I am aware, the aromatic compounds are the only ones to change/react over time – the narlies remain constant. For this reason I’d expect to see only a marginal drop in the measurements as time goes on, certainly not approaching a 2% change.

    It seems to me that aromatic solids, oils and carbon dioxide have a much greater effect on the colour of coffee (especially espresso) than simply the concentration of total dissolved solids.

  4. robert @ elysian

    James…. just a conceptual thing here, but have you considered trying to determine how to accelerate staleness? Could an investigation into this as a process help derive a more objective way to measure it?


  5. AndyS

    robert, I don’t know about accelerating staleness, but I’ve always meant to experiment with storing just-roasted coffee at higher-than-normal temperatures (say 30C) in order to “age” it faster. Will it be ready to make good espresso a day or two sooner than usual? I’m assuming that aging and staling are different stages of the same process.

  6. AndyS

    James, have you seen this?
    Claims to be a method of objectively measuring the “staleness” of brewed coffee (as it sits on a hot plate). A completely different problem from measuring the staleness of coffee beans (and probably much, much easier), but interesting nonetheless.

  7. Mike

    I have a TDS tester you can have if you want it. I am not so sure you will have any meaningful results from it, as the volatile (and hence unmeasured) components will be the first to go.

  8. Ron Sveen


    The research below article should be published very soon. It offers some insights into the difficulties associated with predicting staleness.


    Risks and pitfalls of sensory data analysis for shelf life prediction: data simulation applied to the case of coffee
    LWT – Food Science and Technology,
    S. Guerra, C. Lagazio, L. Manzocco, M. Barnabà and R. Cappuccio

  9. Martin

    I think Ian Clark brings up an important point. Coffee flavor (as all flavor) would typically be 80% aroma and 20% taste. A total dissolved solids (TDS) measurement sounds nice, but the problem with this is that it is typically measured as conductance. This does indeed say something about the presence of ionizable contents but would not pick up changes in the amount and composition of the volatile organic compounds which give rise to the aroma. To investigate this you would have to do use some kind of GC setup (headspace or solid phase micro extraction?) coupled with an MS. And all of this should of course be related to the olfactory detection thresholds as high concentration does not necessarily mean that it actually will influence the overall aroma.

    So although a TDS measurement seems to be a good “rough” guide for extraction (given that you have a well looked after and calibrated TDS meter and always measure at the same temperature), I doubt it’s usefullness for asessing staleness.

  10. Ben Kaminsky

    Let me know when you want to have a go at it and I’ll set the test up on my end with your same parameters. I think we are looking at least two different tests here, as people are basically saying above. One to determine how long well packaged (sealed foil bag, <15 minutes out of the roaster) coffee lasts at a certain yield, and two, plotting those same cups against blind cuppings. Every four days for two weeks, followed by every 7 days for 4 weeks, follewed by every 14 days for 6 weeks. Seems extensive enough for me. Use a Kenya for the test.

Leave A Comment