Read & Interpret Results

Every quantitative survey produces two views of the data: the raw distribution (bar charts of how respondents answered each question) and a story layer that the platform builds automatically to frame the findings.

The header statistics

Before reading results, always check the four statistics at the top of any results page. They tell you how much weight to give the findings.

Statistic	What it means
Respondents	The number of respondents who took this survey (e.g., 10,050)
Universe size	The total population the respondents represent (e.g., 262 million)
Margin of error	The plus-or-minus range on any percentage finding. Also called confidence interval. A margin of error of 2% means a finding of 45% could be anywhere from 43% to 47%.
Design effect	A measure of sampling diversity. A value close to 1.0 means your respondents were drawn from diverse sources; a value of 2.0 means significant clustering; you would need approximately 80% more respondents to recover the statistical power the raw count suggests. The platform targets approximately 1.2.

For a deeper explanation of these statistics, see Statistics.

Meta audience data (the source for segment building) does not cover 100% of the US population. The platform uses ACS/census data to extrapolate from Meta’s coverage to the full population, which is why the universe size can be many times larger than the respondent count.

The distribution view

The distribution shows bar charts for each question. Each bar represents the percentage of respondents who chose each answer option. This is the raw data.

Read distributions looking for:

Large gaps between options: signals of strong preference or rejection.
High skewness, when most answers cluster at one end of a scale. High skewness is often the most informative finding because it points to something strongly felt (positive or negative) that warrants a qualitative follow-up.
Surprising patterns: for example, if a brand that feels premium shows low trust scores, or a budget brand shows high consideration.

The story / narrative layer

The platform automatically generates a narrative that frames the competitive landscape. For a brand track, this might look like:

“Apple is the leader by a significant margin. Samsung is the closest follower. Google and Motorola are significantly behind.”

The leader/follower framing applies across each metric in the brand funnel. Read the narrative first for orientation, then use the distribution to validate and quantify the claims.

The brand funnel

The brand funnel tracks consumer progression through six stages. For a smartphone example:

Stage	What it measures	Example finding
Awareness	Have they heard of the brand?	Near 100% for Apple, Samsung, Google, Motorola in US
Familiarity	Have they used or interacted with it?	Apple and Samsung very high; Google lower
Consideration	Would they seriously consider it next?	Apple leads sharply
Preference	Which do they prefer overall?	Apple dominant in US
Intent	Likely to purchase in next 90 days?	Apple leads; Motorola near zero
Endorsement	How likely to recommend (NPS)?	Apple highest

The funnel is additive and directional; a brand cannot have high consideration without reasonable familiarity. Large drops between stages identify where the brand is losing consumers and where intervention is most valuable.

NPS bands

The platform uses standard Net Promoter Score categorisation on the endorsement question (scale of 1–10):

Score	Category	Meaning
9–10	Loyalist	Will actively recommend. Likely to stay and advocate.
7–8	Passive / dicey	Satisfied but not committed. Will leave for a better offer.
1–6	Detractor	Unhappy. At risk of leaving and may discourage others.

The NPS calculation is: percentage of Loyalists minus percentage of Detractors. A positive NPS means more advocates than critics.

Brand perception metrics

Beyond the funnel, the platform surfaces perception metrics for each brand on five dimensions:

Metric	What it measures
Customer support	Perceived quality of customer service and support
Innovation	How cutting-edge and forward-looking the brand feels
Premium feel	Whether the brand conveys quality and status
Trust	Reliability, data security, honesty
Value for money	Whether the price feels fair relative to the benefit

In the smartphone context: Apple leads on trust (security) and premium feel; Google leads on innovation (driven partly by AI association); Motorola leads on value for money (lower price point).

The mission-outcomes and quadrant view

What comes next

Converse with a Twin: take a high-skewness finding and ask the twin “why”
Worked Example: The GEICO Skewness Loop: an end-to-end flow that starts with survey findings
Statistics: deeper explanation of margin of error, design effect, and universe extrapolation