Lab 01Analytical model
Reputation model
One index that connects conversation, search, content and surveys, and that you can take apart down to the signal that moves it.
Beverages category · Brand A · fictional data
01The problem
One score cannot tell the whole story.
A brand can gain visibility while the perception it generates moves in a different direction. An index alone cannot explain that: we need to identify which dimensions are changing, which indicators explain those movements and what else should be investigated before making decisions.
02What it does
- 01
Structured measurement
Ten submetrics from social listening, search, content and surveys are organized into three components: positioning, perception and consideration.
- 02
Traceable results
Each score can be traced to its input data, its source and the rule that transforms it.
- 03
Three brands, the same methodology
Compare three category brands across an initial measurement and a follow-up, keeping the same weights.
- 04
Weight sensitivity
Adjust each component’s weight from 0 to 5 and separate observed change from methodological change: same data, a different priority.
03How it works
- 01
Gather the signals
Social listening, owned channels, search and surveys. Each signal retains its context: competitive category, historical reference or survey responses.
- 02
Apply a rule to each indicator
Every signal has its own rule and becomes a submetric out of 5: category share, weighted sentiment, survey responses or historical reference.
- 03
Group into components
Each component combines its submetric scores using a weighted average: sum of (score × weight) / sum of weights.
- 04
Calculate the index
The index combines the three component contributions. The model divides the weights by their sum; they do not need to total 100.
- 05
Read the evidence
Compare measurements, investigate which indicators explain the differences and compare conversations and survey answers before deciding.
04Data and limits
A summary, not the whole story.
Illustrative demo with synthetic data; the questions and rules are examples. The index is a weighted model score: it is not a percentage, does not estimate sales, does not establish causes and does not claim universal validation as a reputation measure. Changing weights changes the interpretation of the index, not the brand’s actual reputation.