Brand Awareness Calculator
Unaided and aided awareness from survey counts, with the confidence interval that decides whether a wave-on-wave change is real, and excess share of voice.
Sample 600
Unaided awareness 21%
95% interval 17.9% to 24.4%
margin ± 3.3%
Aided awareness 49%
95% interval 45% to 53%
The gap between them 28% recognise you without naming you
Against last time
previous unaided 18%
change +3.0 points
the margin on this wave ± 3.3%
is it real not demonstrably: the change is inside the noise of two samples this size
Share of voice against share of market
share of voice 14%
share of market 9%
excess share of voice +5.0 points
what the literature says brands spending above their market share tend to gain share, roughly half a point of share a year per ten points of excess
What a tighter margin would need
± 5% 255 respondents ← already there
± 3% 709 respondents
± 2% 1,594 respondents
Unaided awareness is 21% with a margin of ± 3.3% on a sample of 600. Any
movement smaller than that margin is not a movement, and quarterly
awareness reporting on small samples consists largely of reacting to it.
Unaided and aided measure different things. Unaided is whether you come
to mind, which is what actually competes at the moment of purchase.
Aided is recognition from a list, which is much higher, much easier to
move, and much less predictive.
The gap between them is worth watching on its own. A large gap means
people recognise you but do not think of you, which is a salience
problem rather than a reach problem, and more impressions rarely fix it.
Excess share of voice is 5.0 points. The regularity from the advertising
effectiveness literature is that brands whose share of voice exceeds
their share of market tend to gain share, at roughly half a point a year
for every ten points of excess. It is a pattern across many brands
rather than a mechanism in yours.
Awareness is not the objective. It is a proxy for being considered, and
a brand can be widely known and rarely chosen. If the survey is the only
measure, the risk is optimising recognition rather than demand.
Sample composition matters more than sample size. Four hundred
respondents who match the buying population beat four thousand who do
not, and a survey run through your own mailing list measures the people
who already know you.
Question wording moves the answer more than most campaigns do. "Which
brands of X can you think of" and "Have you heard of X" produce
completely different numbers, and changing the wording between waves
makes the trend meaningless.
Output is valid and updates as you type.
Fix the highlighted fields to update the output.
A brand tracker on a sample of 600 measures unaided awareness to within about three points. Most quarter-on-quarter movement in awareness reporting is smaller than that, which means most brand tracker meetings are spent reacting to noise.
So this prints the interval next to the number, and when a previous wave is entered it says whether the change is larger than the combined noise of two samples that size. On the example, 18 percent to 21 percent is not demonstrably a change.
The other distinction it keeps is between unaided and aided. Unaided is whether you come to mind, which is what actually competes at the moment of purchase. Aided is recognition from a list: much higher, much easier to move, much less predictive.
How to use
- Put in how many people were surveyed and how many named you without prompting.
- Add how many recognised you from a list. Aided is always at least the unaided count.
- Enter the previous wave’s unaided figure to test the change, and the share of voice and share of market to see the excess.
Example
600 respondents, 126 unprompted, 294 from a list, against 18 percent last wave:
Unaided awareness 21%
95% interval 17.9% to 24.4%
margin ± 3.3%
Aided awareness 49%
95% interval 45% to 53%
The gap between them 28% recognise you without naming you
Against last time
previous unaided 18%
change +3.0 points
the margin on this wave ± 3.3%
is it real not demonstrably: the change is inside the noise of two samples this size
Share of voice against share of market
share of voice 14%
share of market 9%
excess share of voice +5.0 points
what the literature says brands spending above their market share tend to gain share, roughly half a point of share a year per ten points of excess
What a tighter margin would need
± 5% 255 respondents ← already there
± 3% 709 respondents
± 2% 1,594 respondentsHalving the margin costs roughly four times the sample. That is the arithmetic behind every tracker budget conversation.
Pitfalls
Read the margin before the change. Three points of movement on 600 respondents is what the same brand would produce twice in a row with nothing happening in between.
Two waves have more noise than one. Comparing wave to wave combines both margins, so the change has to clear roughly 1.4 times a single margin before it means anything. The tool applies that rather than comparing against the one-wave figure.
The gap between aided and unaided is its own signal. A large gap means people recognise you but do not think of you, which is a salience problem rather than a reach problem. More impressions rarely fix it.
Composition beats size. Four hundred respondents who match the buying population beat four thousand who do not. A survey run through your own mailing list measures the people who already know you, and it will produce cheerful, useless numbers.
Wording moves the answer more than campaigns do. “Which brands of X can you think of” and “Have you heard of X” are different questions with different answers. Change the wording between waves and the trend is gone.
Awareness is not the objective. It is a proxy for being considered, and a brand can be widely known and rarely chosen. If the tracker is the only measure, the risk is optimising recognition instead of demand.
Excess share of voice is a pattern, not a mechanism. The half-point-a-year rule comes from aggregate advertising effectiveness research across many brands. It describes a tendency in a population, not a promise about yours.
Compatibility
Arithmetic in the browser: nothing is uploaded and nothing is stored.
The intervals are Wilson score intervals rather than the normal approximation, which is why a count of zero produces 0% to 0.6% instead of a symmetric interval that runs below zero. The test suite asserts both boundaries, since a tracker that reports a negative lower bound on a low-awareness brand is the usual sign of the wrong formula.
The change test compares the absolute movement against the single-wave margin multiplied by the square root of two, which is the standard error of a difference between two independent proportions. It is an approximation and it is the right one at these sample sizes.
Sample sizes for a target margin are computed at 95 percent confidence using the observed rate rather than the conservative 50 percent assumption, so they are smaller than a generic sample size table would suggest, and they are correct for a brand at this level of awareness.