« Back to Maxwell’s Blog

Dont Change a Thing - The Set Recipe Ideal

April 04, 2013

Choose a fixed recipe, dial in to that recipe and stop there. Bingo. Done. This will display the best of the coffee and, if for some reason it doesn’t taste right then the coffee must be faulty, because of the roaster or something else that happened elsewhere. The barista’s responsibility has nicely been absolved. Yes, under these circumstances (the set recipe ideal) there is a responsibility to brew to a specific recipe and meet the numbers, but there is not a relationship with the numbers that is explicitly taste driven (what I mean by taste driven is using one’s palate to diagnose flavour issues/attributes). Where do these set numbers come from? Who chose these numbers and how were they chosen? What are they built upon? Do they do the complexity of coffee justice? Are they fully representative of the reality of coffee in its breadth and entirety?

Now the set recipe idea would seem to mean different things for different people, For some it is brewing espresso to 3 set numbers - Dose weight, Shot weight and shot time i.e. 20g, 30g, 28sec. For others it may be a ratio and an extraction yield, i.e. a 65 percent brew ratio to an 18 percent extraction. Here you would have to change grind and extraction time to achieve the extraction percentage. Either way it’s a set of goal posts from which flavour follows as opposed to flavour evaluation that is followed by goalposts.

I don’t think it’s the perfect solution, or even a very good one. In many a case though, the notion of choosing a set of numbers and sticking to them is taking hold. None of my experience with coffee has ever suggested this and I am not happy serving coffee crafted in this way.

Everything I have observed and learned about coffee presents a more complicated and variable reality. I treat the brewing role as a taste/flavour driven one, aiming to problem solve, changing parameters to create recipes that work the best. I appreciate that there may not be only one recipe and that any exploration has its limits. However a set recipe is the starting point from which a net can be cast (variable recipes). This then allows a larger view of the coffee and the chosen recipe can then reflect the chosen view, i.e the best bits. It simply allows more sensory data input, which then allows for a brewing approach that mimics and empathises with the variable and complex world of coffee. It’s a taste survey with more data, with more breadth.

Before this becomes a polemic pro or anti numbers in coffee debate, I would like to say that I am highly interested in any data or observation that helps record and log our tasting experiences as well as that which may allow us insight, leading to better brewing quality in the future. But I will only use them to dictate my brewing if the results are conclusive, if they result in the most positive displays of coffees and on top of that on day to day basis.

First of all, I would like to look at the wider world of coffee and forget about the role of the barista for a moment. It is nice to do this, it adds perspective.

Straight behind the barista role is that of the roaster. I phoned around before writing this, just to check on something. A lot of the time I felt like I was asking a really stupid question, it was often greeted with a little silence and confusion. Q. “Do you roast all of your coffee to the same profile, regardless of origin, age or anything. Do you just treat all coffee like it’s the same simple thing?”

No was the answer. Yes, all of them stated that they had a starting point based on the coffee and …… wait, stop, we are already in a position where the differences are being highly appreciated and monitored. It doesn’t stop here though. These starting points have been built off of experience but the roasters are well aware that it doesn’t guarantee the best result. They cup the coffees and change or develop the profiles based on taste diagnosis. Yes they all will have different procedures and different methodology, but there is a thread of assessment which leads to positively and purposefully changing the recipes to achieve something greater. One roaster quite rightly related this whole topic to sample roasting. He has always wondered about the restrictive nature of small sample roasts. Were the coffees he chose just the ones that best suited the sample roast profile? Now, defects and problems will still be spotted in the sample roast, but the difference between two good coffees and the one he chooses to work with, that is the question. Has he really seen what the coffee has to offer? This quickly brings about the question of can we ever see all of what a coffee has to offer? The fact that the answer is no does not then logically lead to the solution of sticking to just one recipe.

Very quickly the role of the roaster is used to justify the rigidity of a never changing brewing approach. The suggestion is that the roaster should have tackled all of the variables so that the barista just has to brew the coffee to a standardized recipe.

Poor old roasters, the blame is shifting back across. Their product is once again being poorly utilised and restricted, although this time not because of limited ability to control the brew (i.e. brewing shots to sight and not appreciating grind or dose changes). This time it’s because the solutions that have meant for more repeatable brewing have supported an emboldened notions that the brewer no longer needs to take part in taste direction, that all has been fixed.

I am not suggesting  a shift of responsibility straight back onto the brewer, it’s a shared responsibility. It may not be the clean simple answer that some coffee people are looking for, but it’s a reality. I have always felt a responsibility as a brewer to utilise the raw ingredient. Now, I am not sure if the one brew recipe approach is implying that if baristas take the mantle of responsibility too heavily then roasters and growers won’t develop. I’m not sure if anyone actually thinks this, if they do it is ridiculous.

The other end of this same point is the suggestion that if there is a set end point i.e. a set recipe and extraction percentage, then everyone else in the coffee chain can work towards it. This shift of responsibility may not be seen as one but it is. Does this (fairytale) ideology really fairly reflect the nature of this complex agricultural product? Is it really the next logical step based on all of the tasting experience that coffee has granted us?

It would seem that in some ways there is a regression back to god shot ideologies. Maybe we could run a shop where we advertise that we always brew in the same way. This allows us to showcase the incredibly changeable nature of coffee – that some days it’s sharp and thin and that other days it’s sweet and structured, and all at the same recipe! How interesting and marvellous! Well interesting yes, but marvellous, no.

Now I would love it if set recipes delivered the best results and continued to spotlight individual coffees and roasters. I am not anti-systems, or anti-numbers if they achieve consistent quality. It seems to be a blind belief though, and like all blind beliefs, the evidence is easily ignored. The evidence in coffee is flavour. Yes this is a difficult arena to deal in because all sorts of biases and variability/subjectivity come in to play. But the set recipes themselves have been informed at some point by flavour to attain some averages, except now they are not averages, they are a definitive singular correctness.
So where have these “perfect” recipes come from? These perfect numbers?

They have come from averages based on evidence, which evidence? Taste evidence. A Jenga tower built upon various inputs. Refractometer suggestions are themselves rightly broad, the difference between the same weight of dose and water at the same temp, but with different grind to achieve an 18% extraction, is staggeringly different to the same recipe with a 22 % extraction, and those behind VST have not suggested one set recipe but instead have made some observations and displayed the results. Observations derived by people tasting. It is the way this information has been snatched and run away with that is quite astounding.

As far as my understanding of refractometers goes, the equations are themselves built upon averages, different compounds make light refract differently, and light is the measuring device.
I find it an interesting observational tool that, in addition with other data, most importantly that of taste diagnosis, can present us with a more specific curiosity. It also has the less desirable ability to stop a better taste goal being achieved if the numbers seem to lie outside of the accepted range. “Whoa, this coffee tastes under extracted but we are already at 21 percent, we can’t go much further.” Maybe. Maybe not.

I do find informed, accurate logging as we build recipes to be a necessity: weighing dose, water and monitoring time and temp, all of these help us understand what the coffee we just made was, and we can relate it to taste at that moment in time with those variables at that time. This is not the same as making every other coffee ever to those same numbers. Personally, I follow the coffee until I’m happy. Now and then this can sit at 23 percent or 17 percent. Overall though, I try and constantly be open minded as to how far we stray from our anchor points. The Rwanda espresso recipe at the bottom of the page is an example of going to a lower brew ratio that is outside of the normal variance to achieve a desirable result.

I also appreciate all the other issues of taking part in flavour direction, as I mentioned in the prism of the barista. It necessitates baristas who make judgement calls rather than just being a bystander. Now I don’t have an issue being a bystander (if it works). I got into coffee because of the flavour possibility and I always intend to embrace craft and method wherever it aids this display of flavour. However much we want coffee to be like a bottle of wine, currently it just isn’t even close to a finished product when it reaches the brewer.

Another possible concern here(and the reason set recipes seem desirable) is that the brewer can otherwise get in the way by pulling recipes that don’t reflect the cup. There’s a problem here though – by keeping constant and doing nothing, you are still having an effect, its effect through neglect. It may seem like a cosy little thought “I keep everything the same and all the changes are issues elsewhere”, but now we are heading back to the romanticising of variability at the cost of the quality of what’s being served. As far as I see it, if I can have a positive effect then I will take part in the process and I will encourage all of my staff to do so. This is more difficult to scale, and it requires more experienced, taste driven baristas and more communication, but I think it is worth it. I think the results are worth it.

I recently saw someone on Twitter compare brewing to control theory. I wasn’t aware of this, but for me it’s a dead ringer for coffee brewing.

“Control theory is an interdisciplinary branch of engineering and mathematics that deals with the behaviour of dynamical systems with inputs. The external input of a system is called the reference. When one or more output variables of a system need to follow a certain reference over time, a controller manipulates the inputs to a system to obtain the desired effect on the output of the system.
The usual objective of a control theory is to calculate solutions for the proper corrective action from the controller that result in system stability, that is, the system will hold the set point and not oscillate around it.”


Conceptually this could be applied to both the set recipe approach and the variable recipe approach. If you intend your output to be a set recipe then the controller must manipulate changes in grind and dose etc. to realise the set recipe. But let’s change that goal. Let’s make the goal a flavour output. A positive taste balance and structure. When this is the output, the controller then needs to change the parameters of the recipe that dictate how the coffee tastes in order to achieve a set output of positive flavour quality. Once this is achieved we need to employ the previous system of achieving that chosen recipe, it’s as if there are two control systems operating simultaneously.

This whole topic is a two pronged discussion. There is the argument that the recipe should stay the same and then quality will shine through, but that taste problems are frequent. The deduction is then that there has to be a fault elsewhere (as stated, I disagree with this). The other prong is that that there isn’t variability at all, and that this approach does actually achieve consistent flavour results(that means not just with the same coffee but across all good coffees). The first side is one I can understand, as we are both viewing the same results, the second is going to be bloody hard to reconcile as we just disagree on the quality of the output.

The first viewpoint has several problems, especially when many an individual has told me they are not happy with taste results quite often. Either they find that as the coffee changes the flavour isn’t as good, or other changes mean the results are not pleasing, and I don’t just mean slightly different, I mean objectively bad. Maybe it’s partly the change in water supply. It doesn’t matter whether you have reverse osmosis or a filter with magnesium resins, the water going in to most cities and towns/areas is variable – different reservoirs, different processing.

We have noticed huge variances in our water, not just in TDS (total dissolved solids) but in pH level. It’s acidic one week, alkaline the next. After chatting to some water people, it seems it would be possible to use reverse osmosis to go down to distilled water and then re-mineralise at a set rate, but I’m not aware of anyone in speciality coffee doing this, the people complaining about taste definitely aren’t. Word is Starbucks are giving it a go. Secondly it appears that many of the changes in coffee making variables do not have a linear relationship, so a change in pH or magnesium in the water is not likely to have the same degree of change in flavour for two different coffees.

Anyway, even if a set water content could be achieved, we still cannot control ageing from roast and the variance in the green bean*. We still can’t and will never be able to control the sun at the farm, the changeability of terroir. Yes, we will work on things we know have a positive impact, eradicating defect, picking ripe, etc. We can all have an effect, hopefully if we gather data and carefully move forward we can have a scalable effect (but I don’t see how it can ever be more than a calculated estimate driven scaleability). We may know that two lots of naturally processed coffee will taste different – one with hourly turnings vs one with turnings every two hours. But we will only know exactly how different they are when we taste them. The one step recipe ideal is a decision not to try and take part in improving and developing,

There have been lots of improvements in the understanding of coffee, they are all based upon trying to understand why things happen, how coffee works. I am worried at the suggestion that complete and singular markers have been achieved. Each individual can brew how they wish but I wanted to use this blog to explain why we don't brew to set recipes and what I think all the drawbacks are.

Here is an example of the recipes we are using as I write this.

We use a Strada and a chosen pressure profile.

Mojo is done after recipe is reached through taste diagnosis

1 - Colombia - Finca Potosi - Pulped roasted by Origin
18.5g dose to a 33g shot in 29 seconds extraction – 17.7% extraction 10% TDS
The Origin coffees are roasted on a Loring Smartroast and for some reason we find that their coffees develop the most positive flavour when passing the 14 days from roast mark (they have a grassiness and astringency when close to roast, regardless of recipe). We therefore manage our stock from them so that it does not go on before this point. We have changed the recipe once since having this coffee on over the past week. We started with 19g to the same 33g but have dropped the dose as that recipe lost positive balance.

2- Birembo - Rwanda - washed - roasted by Hasbean
18g dose to a 37g shot in 30 seconds – 20% Extraction 10% TDS
This is a particularly low brew ratio for us. Interestingly it lacked sweetness and cleanness at a higher brew ratio (regardless of other parameters) It was roastier at the higher ratio. I see it a lot like putting salt on a grapefruit, it tastes sweeter, not because you have added sugar, but because of the interplay of flavours and their relative perception- maybe

3 -Dibello -Ethiopia - Washed roasted by Tate.

19g dose to a 28gram shot in 31 seconds – 17.7%extraction 11.7 TDS
This coffee is currently 14 days from roast, and we have not had it on for more than two days. We change each espresso every 4 days, so by aging it to a reasonable degree in the first place we can develop a recipe based on taste diagnosis for each coffee and see little negative change, although not always. We taste check morning and mid-day.

We have often wondered if it’s the TDS and mouth-feel that allows espresso to taste more balanced at a lower extraction, although these numbers don’t really answer this. They all taste good though.

It is interesting to get the same coffees back from different roast batches/same roast profile, and play around with a longer time arc noting recipe changes, not always though.

I do think the variance and impact of recipe is slighter than espresso but we approach it in the same way and find it important also.

We start with 15g to 210 water with a starting grind and then alter dose and grind to taste whilst keeping water weight constant. We Mojo afterwards.

Aeropress - Passeio, Brazil, Roasted by James Gourmet Roasted 12 days ago
15g dose to 210g of water, starting at 94 degrees, grind setting 3 (this isn’t really a very useful piece of information but rather a reference point for the filters made here on the same grinder)extraction percentage of 20.5%.

1 Yirgacheffe Wote - Natural, Ethiopia from Hasbean, roasted 9 days ago.
15g dose to 210g water at 93 degrees, 1min 15sec brew time once up top at a grind setting of 3, Extraction yield of 18.7%.
2. Passeio pulped, Brazil roasted by James Gourmet coffee 12 days ago.
15 g dose to 210g water at 93 degrees for 1min 15sec brew time once up top at a grind setting of 2.75 with an extraction of 21.5%.

Clever Dripper
Pixcaya, Guatemala, washed, roasted by James gourmet 8 days ago.
15.5 g dose to 210g water 2 minutes immersion, starting water temp of 94 degrees, on a grind setting of 5. Extraction of 19.5%.

I believe that together as a whole we reach the highest heights, and from where I am standing the classic coffee chain is one where each link is endeavouring to explore the responsibility of how we can process taste information to then adapt and change in order to achieve the most we can. There is a real bonus here as well, it keeps the speciality coffee field as a flavour driven one as opposed to a theoretical one.

And on that note, In contrast to the title of this post, do have a taste and do change a thing or two. Think about which of the flavours in that cup are due to the farmer, which are due to the roaster and which are due to you, the brewer.

P.S. don’t forget to wonder about the other variables, equipment water etc…. ;) Its a lot of wondering. Its a pretty amazing thing really - coffee.



*on this note should the roaster just stick to one profile as the green ages? This isn’t a linear ageing process either, different chemical make ups will respond differently to age, let alone roasting and brewing. Or should they work with the product noting that the original profile relied on the state of the green at that moment in time? Following a set recipe – set roast profile, they could just blame the variance in the coffee and make no effort to maintain a quality they are happy with (I don’t mean past crop ageing) or they could just ignore the changes altogether…


Explore Our Beans

Beans An icon showing a coffee bean.


Environments for Barista Nourishment

March 22, 2013


Taste, Flavour, and Perception

May 04, 2013