You Standardised Everything — So Why Doesn’t It Taste the Same?
You standardised the beans.
You standardised the recipe.
You even standardised the machine.
Yet Site A tastes different from Site B.
In multi-site hospitality, this is one of the most common (and most misunderstood) causes of inconsistency:
The water isn’t the same.
Municipal water changes by location — and espresso reacts fast
Municipal water chemistry varies region to region. Even within the same city, supply sources and treatment can shift.
The key variables that move are:
- Hardness
- Alkalinity
- Mineral content
Espresso is sensitive. Small changes in these inputs can translate into big changes in flavour and performance.
What multi-site drift looks like in real life
Across an estate, the pattern often looks like this:
- One site extracts slightly bitter
- Another tastes flat
- A third scales faster
- Teams start adjusting grind to compensate
- Recipes slowly drift to “make it work”
Over time, your “standard programme” becomes a set of local workarounds.
The hardest part?
The real issue isn’t visible. Everything looks correct on the surface: same coffee, same machine, same training… yet different results.
Why grind adjustments don’t fix the underlying problem
When water is different, grind changes are usually just compensation.
That compensation:
- masks the true cause
- increases variability between shifts
- makes training harder
- turns QA into guesswork
You end up chasing a moving target — and blaming the coffee, the barista, or the machine.
Water isn’t a background input — it’s a controlled variable
In multi-site operations, water must be treated like any other key input you control.
Not background. Not optional. Not “nice if we get to it.”
A controlled variable means:
- you measure it
- you define a standard
- you manage it site-by-site
- you hold it steady over time
The blunt truth: without measurement + management, consistency is impossible
If water isn’t measured and managed per site, your consistency challenge isn’t a “training problem”.
It’s a maths problem.
Different inputs produce different outputs — even when everything else is standardised.
Practical next steps for multi-site teams
If you want a programme that truly behaves like a programme:
- Baseline each site (don’t assume “filtered” = same)
- Set a target standard for coffee water
- Match treatment to the site (starting water differs)
- Build water checks into QA (like recipe and calibration)
- Track scale risk + maintenance patterns to catch drift early
When water is controlled, beans and recipes finally behave the way you expected them to.
Alnur Merali
Nairobi Coffee and Tea Company
