Accuracy vs Precision

The dartboard analogy

Imagine two dart players, A and B:

• Player A throws darts that land all over the board but cluster near the bullseye. His average score is high; he is hitting the center area consistently enough to rack up points. He is accurate but not precise: the darts are scattered across a wide area even though the center of that scatter is the right place.

• Player B throws every single dart into the exact same spot, the top-right corner of the board, far from the bullseye. His individual darts land in a tight cluster: nearly zero variance between throws. He is precise but not accurate: he hits consistently, but not the right target.

The question is: if you had to coach one of these players to improve, which one is the easier job?

Player B, the precise one. All you have to do is help him aim at the center. Once he locks in the right target, his precision means he will hit the bullseye consistently. Player A is harder to coach because you do not know where the next dart will land. There is no single thing to fix.

The unit of precision: variance

Variance is the statistical measure of how spread out a set of values is. The smaller the variance, the more tightly clustered the measurements, and the more precise.

• High variance → low precision → measurements scattered widely
• Low variance → high precision → measurements tightly grouped

Accuracy and precision are independent. A measurement system can be:

	Low variance (precise)	High variance (imprecise)
Near true value (accurate)	Ideal	Lucky on average, unreliable each time
Far from true value (inaccurate)	Consistently wrong; easiest to fix	Scattered and wrong

Deterministic vs probabilistic: where variance actually lives

Variance is a property of delivery, not prediction. When you say “I think there is a 25% chance this person is 25 years old,” you are describing the probability of a prediction being correct. That is uncertainty in prediction; it is not the same thing as variance. Variance requires multiple data points and measures how spread out those actual measurements are.

In marketing research, variance appears in how a measurement behaves across repeated observations: how consistent the signals from an audience are when you query them at different times, for example. It is a hard measure of real-world consistency, not a soft expression of how confident you feel.

Why marketing is about direction, not accuracy

A fundamental principle at Consumr.AI: marketing is never about accuracy; it is always about direction. As long as you have precision, consistent signals pointing in the same direction, you can make a decision.

You do not need to know that exactly 47.3% of a target audience prefers a particular message. You need to know reliably that they lean toward it, and that the lean is consistent enough to act on. Precision is what makes that confidence possible. Accuracy (in the strict statistical sense) is less critical because you are rarely measuring an absolute ground truth; you are reading consumer signals to make a directional call.

Practical application in Consumr.AI outputs

When you look at results from a behavior report or survey, ask:

• Are the signals consistent across different question angles? (Precision)
• Are they pointing in the right direction for the client’s objective? (Accuracy)

A high-variance result, where different questions or different runs of the same analysis point in opposite directions, is a flag worth investigating. A tightly clustered set of signals, even if they are not perfectly “correct” in an absolute sense, gives you enough to recommend action.

See How to Read and Interpret Results for how these concepts translate into reading platform outputs.