Can anyone ‘analyse’ your house coffee blend?

This article has also appeared today on the news portal

A coffee roaster has spoken this month of his catering trade customers being approached by rivals with an extremely bizarre sales tactic – and now the matter has come to light, other coffee roasters have spoken of experiencing the same tactic from competitor salesmen. It involves the alleged ‘anaylsis’ of a roasted coffee blend.

(For obvious reasons, in this story we have avoided the names of the main participants)

A well-established coffee roaster reports that he was supplying a restaurant with a hundred-per-cent Arabica house blend. A rival came along touting for the business, and said to the client: ‘I think your supplier is telling you fibs about what he’s selling you – it isn’t hundred-per-cent Arabica. Give me a sample of your house blend, and I’ll have it analysed’.

The caterer gave him a sample, and it was taken away. In due course, the salesman came back with information.

The result was actually given in writing, and was phrased obliquely. The would-be supplier reported that the existing blend ‘would appear to be… smooth, even, light flavour, little flavour’, and while not this time actually referring again to the promised ‘analysis’, suggested that ‘we expect the following coffees to provide a close match…’

The following suggested blend involved thirty per cent Vietnamese robusta, and the rest of it was x per cent of a certain origin, y per cent of another origin, and z per cent of a third origin.

The client was unhappy. Although the would-be supplier had avoided saying that the promised ‘analysis’ had been done, the proposal certainly inferred that it had, and it gave the client considerable cause for concern – because the clear suggestion was that his existing blend involved a large robusta content. If he were found to have been advertising hundred-per-cent Arabica which actually had some robusta in it, he could be in trouble with the trading standards people.

The client confronted his existing roaster, who, equally annoyed, went to the trouble of showing the client his recipe, proving that the ‘analysis’ was wrong in every single respect… not one ingredient was right. No robusta was involved, and some of the others ‘identified’ were arguably of extremely ordinary provenance, compared to the quality of the Arabica which had actually been used.

He then showed the client his sacks, and invited him to watch the blend being roasted.

In this story, it is the question of ‘analysis’ which has exercised various members of the coffee trade this month. Trying to win somebody’s else’s client is an everyday matter, and so are wild statements by coffee salesmen. Trying to ‘match’ an existing blend is an extremely common occurrence. But is it really possible to ‘analyse’ a roasted blend, and tell what went into it? And what is the justification for trying to win business by scaring the customer out of his wits?

The alleged perpetrator of the ‘analysis’ tactic was challenged about this, and has responded through a third party, who reported it thus:

“They were adamant that they only do this on client request, and they were also adamant that the company they send it to for analysis is massive and very reputable, and has been doing it for many years with a high degree of accuracy. They were adamant that this company absolutely can break out the beans that have gone into a blend.”

This claim is so remarkable that many people in the coffee trade were asked for their opinions on it. And many were very quick to give them.

The supplier behind the alleged tactic had actually given the names of two companies to whom they sent coffee for ‘analysis’. Both of those companies were invited to comment, and reported the following (both of them are extremely reputable companies, but again, their identities have not been given).

The first company said: “this is something we don’t know. We assume that it can be done, but have never tried it.” The second said: “can you do a DNA test to tell arabica from robusta? Yes, but it is expensive and not standard practice. I would suggest that one roaster is using an underhand way of gaining business from another roaster.”

In its reference to DNA, this second company alludes to several items of scientific work which have indeed been done.

In the study ‘Compositional Analysis of Coffee Blends by near Infrared Spectroscopy’ (G. Downey and B. Spengler, 1996) the researchers did refer to a DNA difference between Arabica and robusta, said they failed to achieve sufficient discrimination between the two, but ‘saw promise for quantifying the robusta content of mixtures’. In ‘Coffee species and varietal identification’ (by Tornincasa, Furlan, Pallavicini, and Graziosi) it was said that the researchers presented a method to achieve an analysis to show robusta in a blend, making possible the detection of less than five per cent of robusta content.

In the delightfully-titled ‘Mass Spectrometry-based Electronic Nose’ (by De Winne, Van Leuven, and Dirinck) the researchers analysed all Nespresso blends, and while being able to show a recognisable identification of decaffeinated coffee, and being able to distinguish different ‘flavours’ among the blends, did not claim to be able to distinguish exactly what the different ingredients were.

Some specific work in Hawaii, undertaken in an attempt to stop counterfeiting of Kona coffee, has successfully identified ‘pure’ Kona coffee from coffees sold as Kona but from a different origin. Again, however, it was unable to say exactly what the others were.

In all this work, a certain amount of identification was shown to be possible, but not the compositional breakdown of a roasted blend.

What, then, does the coffee trade think of the alleged ‘analysis’ of a blend, and the claim that the contents of a roasted coffee blend can be identified, in detail?

“Some years ago we were approached by a legitimate university to fund research into DNA fingerprinting of coffee species,” says Stephen Hurst of the green-bean importer Mercanta. “I would love to be able to ‘unlock’ a blend and expose the massive frauds that exist, but to do so you would need a DNA profile from seeds and beans from coffee farms all over the world – a nearly impossible task. But some modest version may be able to identify origins – though I doubt this, as seeds often cross borders.

“I am sceptical that anyone using current technology can re-engineer the blend components of roasted coffee.”

Of all the other suppliers in the trade who were asked about this, in a list which included several roasters of extremely high renown and several green bean importers, not one person (except for the perpetrator of the sales tactic) believed that it is possible for the composition of a roasted blend to be scientifically analysed and the ingredients identified.

From Illy, Marco Arrigo did say that he has sent blends to his lab in Italy for analysis, but for different reasons: “when we look at competitors’ coffees, we look at their defects to judge the value and quality of their coffee. All you can hope to identify is pest invasion, production damage, percentage of oxygen in packaging, presence of fungus or mildew, or maybe age between beans in a batch with carbon dating… but that’s too expensive for regular testing.”

Virtually every roaster who responded said that if they are asked to judge, assess, analyse or copy another blend, they do it by the same basic instruments – the senses.

“We would look at the bean appearance, size, shape, centre cut,” remarked Simon Wakefield of the importing company DR Wakefield. “Do sacks come from origin blended as arabica and robusta? Yes, they do! But robusta beans look different to Arabica beans, both in the raw and in the roast. Different Arabicas and varietals also look different from each other.”

When invited to copy an existing blend, as many roasters are asked to do, extremely experienced coffee tasters can often make a very good educated guess as to the likely components of that blend – but that is a different thing from a claimed scientific analysis.

The basic instrument is the best, came the response from Ian Steel of Atkinson’s in Lancaster.

“I have a very sophisticated onboard analsyer tool called ‘The Nose’! It can detect robusta before it even hits the palate. The best noses in the business, like Jeremy Torz of Union Hand-Roasted, can detect the composition of a blend by nose – he did this once with one of ours that I thought was quite complex.”

The compliment was politely deflected by Jeremy Torz himself, who reported that he too has come under attack by the same tactic of supposed ‘analysis’.

“The same thing happened to me a few years back. A client was approached similarly by another company saying the coffee we had supplied was ‘poor’. They also claimed to have had an analysis done and that also came back with duff information.

“The other company was a large multinational. I was told about it, wrote to the CEO, and got a written apology after I threatened to go public and sue for defamation! They claimed that the action was that of their rep and was ‘not company policy’, blah, blah, blah…”

Only one reputable roaster in Britain, an extremely big contract roaster, is known to have referred to analysis of a roasted blend. The founder of that roastery is said to have told a client that he has an instrument like a scanner – beans go on the glass, the lid is put down, and the machine prints out a recipe of the origins, quantities, and how long they were roasted for. However, it is wise to approach this story with caution, because it is also true that this reputable roaster has a very lively sense of humour, even when teasing clients!

In the entire trade, the general opinion has been the same. Analysis of a blend cannot be done. It was universally thought than any caterer approached by a would-be supplier who offers to ‘analyse’ their existing coffee blend should exercise great caution.

And by far the most entertaining response, from the many who commented, was from roaster Steve Leighton, of Has Bean in Stafford:

“I’ve seen this stuff happen before, and it really is sad. I once had a customer who had an ‘Italian’ brand salesman go into his shop and do an ‘analysis’ of his blend by shaking the beans near to his ear… and then saying: ‘this contains a lot of Brazil’!”


The Caffe Culture Portal is a leading online news service for the UK coffee trade. Boughton’s Coffee House is the leading trade magazine for the UK coffee trade.


%d bloggers like this: