Question set
Category, intent, question and set version.
AiVisible proprietary technology
A system that preserves the context behind AI answers needed for decisions.
We capture questions, full answers, models, brands, sources and history, so the analysis does not collapse into hand-picked screenshots.
We describe only capabilities available in the current product version.
Evidence trail
A finding you can trace
Buyer question
„Which companies should be considered in this category?”
1Full answer and model
2Date and question set
3Detected brands and products
4Sources and rationale
5Analyst annotation
Client data is anonymised or published with permission.
Captured data
Every aggregated result leads back to source material.
Evidence trail
From question to finding without losing context
Active system structure
01
Prompt set
category · intent · version
02
Model run
model · date · market
03
Answer
full answer · sources
04
Coding
brands · arguments · type
05
Finding
observation · limitation
Measurement ID
Unique run identifier
Unique run identifier
Analysis-rule version
Criteria used in the measurement
Criteria used in the measurement
Reviewability
Full answer behind every finding
Full answer behind every finding
Answers
Full content, model, date and context.
Competitors
Brands and products appearing in the same sample.
Sources
Domains and material types supporting the answer.
Variability
Variability is part of the data, not an error to hide.
Answers may change by model, date and context. We therefore compare sets and waves instead of promising a fixed position.
results by model
control question core
history of measurement waves
full answer behind each metric
The system's role
Technology gathers evidence. Strategy gives it meaning.
The system does not know customer value, genuine product strengths or implementation capacity. Strategic work contributes that context.
Data flow
From question set to finding without losing context.
1. Sample definition
Category, intent cohort, question, model, market and set version.
2. Answer capture
The full answer with its date and run information.
3. Coding
Brands, products, recommendations, arguments and sources using explicit definitions.
4. Aggregation
Aggregate views remain connected to their underlying answers.
Quality control
The system should make a finding auditable, not merely generate a score.
question-set version
model and run date
full answer behind the result
cohort comparison
observation separated from interpretation
limitations and missing data recorded
Privacy
Client data does not automatically become marketing material.
Access scope, retention and presentation are agreed for each project. Public examples use our own, anonymised or explicitly approved material.
Important:The technology does not automatically ingest confidential organisational knowledge. Client-provided material is used only for the agreed scope.
Next step
Assess a category using evidence that can be compared later.
We begin by defining the buying situation and a useful measurement scope.
FAQ
Questions before you decide
Does the system show one global brand position?
No. Results are analysed for specific questions, cohorts, models, markets and dates. One position would hide too much context.
Do you store full answers?
Yes. Within the agreed project scope, the full answer is the source material for coding and interpretation.
Does the technology automatically decide what to implement?
No. It structures observations. Prioritisation also requires customer value, offer strengths, cost and organisational capacity.
Can results be compared over time?
Yes, when the control sample remains stable and changes in models, questions and research conditions are documented.
Related resources