The number on the front of the bag is the only thing on the pack that looks like it was measured. Everything else is adjectives. So people use it, and they are being sensible when they do: a 5 ought to be more of something than a 3. The awkward part is that nobody outside the company printing the bag has agreed what the something is.
That number is a roast-and-blend index, set by the roaster and checked by nobody. It does not report caffeine content, and it is not strength in the sense anyone measuring coffee uses the word. Two bags both marked 5, from two different companies, are not making the same claim.
Which does not make it useless. It is a rough signal of how dark the roast is and how much robusta is in the blend, and both of those change the cup in ways worth understanding. It just does not mean what the shelf implies.
Strength already has a definition, and this is not it
In brewing, strength is concentration. It describes how much dissolved coffee is carried in the liquid, and it is measured as total dissolved solids. Jessica Easto's glossary in How to Taste Coffee defines it exactly that way, and she makes the point that strength travels with body: a more concentrated brew feels heavier in the mouth, a weaker one feels thin.
The thing to notice is when that number gets decided. It is set at the moment you brew, by how much coffee you put in and how much water you run through it. The same bag will give you a cup at half the concentration or double it depending on nothing but your dose, which is why one ratio covers every filter brewer you own and why that ratio is where the control actually lives. Concentration is also a separate axis from how much of the coffee you dissolved, and running those two together is behind most of the cups that taste wrong for reasons the drinker cannot name. We took that apart in strength and extraction are two different measurements.
A printed number cannot state strength. The bag has no idea what your dose is going to be.
What the number is actually indexing
Two things, mostly. How dark the coffee is roasted, and how much robusta is in it.
Robusta, Coffea canephora, carries more caffeine and more naturally bitter compounds than arabica. How much more is disputed even among the people who write books about it. Gavin Fridell puts it at roughly 50 percent more, Jonathan Morris at about double. Neither figure comes from a measurement study and neither author offers it as one, so treat both as trade shorthand rather than a number anyone could count with.
Morris also supplies the history that makes a high number legible. Dark roasting is partly a technique for managing robusta. By 1960 around three quarters of French coffee was robusta, and heavy caramelised roasting was the tool used to work against the bitterness those beans brought. Dutch and Belgian preferences follow the same colonial supply lines, and southern Italy's taste for robusta began as a cost decision before it hardened into a palate. National dark-roast traditions are, in large part, supply chains that turned into preferences. The number on the bag is the most recent layer of that.
Why it cannot be a caffeine reading
Caffeine is close to thermally stable through roasting. What changes is the bean around it. Roasting drives off water and volatiles, and Racineux and Tran, writing a roasting guide rather than a caffeine study, put the loss at 11 to 22 percent of the bean's mass. Per gram, that arithmetic leaves a dark roast marginally more caffeinated than the same coffee taken lighter. Per scoop it is less caffeinated, because the darker bean is less dense and fewer grams fit in the scoop.
Both sides of the café argument are therefore correct and neither has stated its units. This is one of several reasons a scoop is a volume measure pretending to be a weight, and it is why a caffeine claim made on the front of a bag is not checkable even in principle.
The measured variance makes the point harder. Crozier and colleagues, publishing in Food and Function in 2012, ran HPLC analysis on 20 commercial espressos bought in a single city. The results ranged from 51 mg to 322 mg of caffeine per serving, with a mean near 110 mg. Six-fold, in one city, in cups sold as the same drink. The drivers they identified were the dose of ground coffee, the arabica-to-robusta ratio, the roast and the grind.
For scale on those numbers: the European Food Safety Authority's 2015 scientific opinion on caffeine gives 400 mg per day, and 200 mg in a single dose, as amounts that do not raise safety concerns for healthy adults. The US Food and Drug Administration cites the same 400 mg daily figure. Those are population-level figures from named bodies and they are not a limit for any particular person. If you are counting caffeine because a doctor asked you to, a scale printed on a bag is not a measurement, and the person who asked you to count is the one to ask what the count should be.
What darker actually does to the cup
This is the part the scale gets most wrong, because the effects are not one-directional. Kenneth Davids' account in Home Coffee Roasting is the clearest available, and it runs like this.
Acidity falls steadily as the roast goes darker. A green coffee with sharp acidity does not lose it cleanly, though. It can resurface at the dark end as pungency, which is a different and rougher sensation.
Body rises as the roast approaches medium-dark, and the mouthfeel gets rounder and fatter with it. Then it reverses. Pushed into a very dark roast, body falls again and the mouthfeel goes lean and gritty. Aroma follows a similar arc, at its most developed through medium to medium-dark and simplifying as the roast continues past that. Origin character shows best from light to medium, becomes progressively harder to find as the roast darkens, and is effectively undetectable at the darkest levels.
Read that against a five-point scale and the problem is obvious. If a 5 on a given brand's scale means very dark, then a 5 can deliver less weight in the mouth than a 3 from the same range while tasting considerably more aggressive. Intensity of impression and body part company at the dark end, and the scale has no way to say so.
There is one more effect worth naming. Underdeveloped beans that stay pale through roasting are visible in a light roast and invisible in a dark one. Dark roasting is a real tool for getting consistency out of variable green coffee, which is a legitimate job. It is also, unavoidably, a good place to put things you would rather the buyer did not see.
What this will not do
Understanding the number will not let you compare two brands with it. The scales are not calibrated against each other and they are not even the same length: as of August 2026 you can walk one aisle and find bags scaled to five sitting next to capsule boxes scaled to thirteen. A 4 and a 9 may describe the same roast. Nothing on either pack lets you find out.
It will not tell you whether the coffee is any good. No number on the front of any bag does that. The useful information is on the back, and it is the roast date. A bag carrying a best-before date eighteen months out and no roast date is telling you what the company thinks freshness is worth.
And it will not tell you that dark blends are bad coffee, because plenty of them are made carefully and drunk happily by people who have no particular interest in origin character. A well-made dark roast that suits your milk and your morning is not a mistake. The claim here is narrower: the number does not carry the information the shelf implies it carries, and buying by it means buying blind while feeling informed.
One last thing it will not do, and neither will this article: stand in for medical advice. The caffeine figures above are population-level guidance published by named bodies, current as of August 2026, and they are not a limit set for any particular person.
What to read instead
Four things on the back of the pack are worth more than the number on the front, and they run in this order. The roast date. Whether the bag names its species or admits to being a blend. The origin, specified to something narrower than a continent. The process, if they bother to print it at all.
If a company will tell you a number between one and five and nothing else, that is not a shortage of space on the packaging. It is a decision about what you are supposed to know. Read it as one.
Where the number does earn its place is as a rough proxy for roast colour within a single brand's own range. If you have found that a particular company's 3 suits you and their 5 does not, that comparison holds, because it is one roaster's scale applied to one roaster's coffee. Carry it across the aisle and it stops meaning anything at all. When the coffee tastes off and you are trying to work out which direction to move, the number will not help, and the sour-or-bitter diagnostic will.