Small trials, fast lessons: how we decide

At Color of Vietnam an idea does not go straight from the meeting room to the shelf. It goes through a small trial built around one clear question. We look at the result, keep what already works, fix what does not, and only then make it bigger. This is our value Small Trials, and this article describes how it runs during an ordinary working week.

Why small first

Because the team is small and the product is food. A full production run with the wrong flavour wastes ingredients, printed packaging and a production slot at a partner facility. A content campaign in the wrong voice takes a brand weeks to walk back. Small trials let us be wrong where it is cheap and fix things where it is cheap.

They are also the fairest way to settle an argument. When two people disagree about a flavour, the quickest route is not whoever speaks loudest. It is two samples on the table, tasted together, with notes written down.

What a small trial needs

Doing something on a small scale is not the same as running a small trial. A useful trial has five parts:

• One question, short enough to write on a single line. For example: if we reduce sugar to this level, does the cracker stay crisp.
• One measure agreed before we start: the team's tasting scores, how often people reach for a second piece, or simply yes or no.
• The smallest scale that can still teach us something. A trial batch at a partner facility, not a production run.
• A deadline. A trial with no end date never produces a conclusion.
• One person responsible for recording the result and proposing the next step.

Without a question, we finish the trial and cannot say what we learned. Without a measure, everyone remembers a different result. We see these two mistakes far more often than the mistake of testing too big.

An example: trying more cacao in an energy bar

Suppose the team wants a PLAY Nutrition energy bar to taste more clearly of cacao. The big idea would be to rework the recipe. The small trial is narrower. The question is written down: if we raise cacao one step above the current version, does the bitterness cover the cashew.

We send the flavour direction to our partner facilities and ask for a trial batch with two samples, the current version and the higher-cacao version. When the samples arrive, the team tastes blind, without knowing which is which. Everyone writes three lines: sweetness, bitterness, aftertaste. Only then do we turn the labels over.

The result is one of three. The new version is clearly better: keep it and move on to a shelf-life check. The new version is harshly bitter: drop it and note the level so nobody tries it again. Or the room splits: not clear yet, try one level in between. All three are good outcomes, because all three answer the original question at very little cost.

Deciding what to keep and what to drop

After each trial we hold a short meeting and settle three things. What stays as it is. What gets changed. What is dropped for good. Everything goes into the sample log with a date and the name of the person who decided. That log matters more than anyone's memory. Six months later, when someone proposes the same idea again, we open the log and see at once where we have already been.

There is a familiar temptation. A trial comes back with a no, but the person who suggested it is attached to the idea and keeps asking for one more round. We handle it with a question of our own: what is the new question. If it is the old question with fresh hope, we stop.

When to scale

We scale an idea when three things are true. The trial says yes. The result repeats at least once more, because one good batch can be luck. And the cost of scaling has been worked out, including ingredients, new printed packaging, the facility's schedule and the team's time.

Scaling also does not mean going all in. We usually widen in steps: a trial batch, a small batch for buyer samples, then the first production run. Each step is a new small trial with a new question. At the buyer-sample step the question is no longer whether the flavour is good. It is whether the buyer reorders.

This way of working feels slow at each step and turns out fast overall, because we rarely have to go back to the beginning.

If you prefer to start with a small question on the table rather than a thick plan, have a look at the open roles on our Careers page. We are looking for people who can frame a clear question, and who know when to stop because it has been answered.