Methodology

How IndexResearch runs comparisons

We start with a real user or business scenario. Before freezing the methodology, we test whether the selected criteria actually measure that scenario. Final scoring begins only after calibration and research-design review.

Research process

1. Research question and sample

We define the audience, scenario, unit of comparison, data cutoff and participant inclusion rules.

2. Calibration

We run trial coding and check comparability, double counting and model sensitivity.

3. Construct validity

We test whether the criteria and weights measure the stated user choice rather than a neighboring capability.

4. Strategic review

Before the public freeze, we assess the information asset the study will create and decide whether to continue, redesign, change format, hold or stop the release.

5. Freeze and scoring

After the model is frozen, weights and rubrics are not changed to reorder participants. The calculation must remain reproducible.

6. QA and publication

We verify facts, links, relationship disclosures and consistency across the README, JSON, the website and other public surfaces.

Ratings and comparative research

The public methodology covers sampling, criteria, weights, evidence, missing data, calibration, construct validity, conflicts of interest, sponsorship, editorial claims and publication decisions.

Open the methodology on GitHub

AI visibility

A separate methodology defines the query panel, AI engine selection, baseline and repeat measurements, mentions, recommendations, citations, Share of Voice and BMR.

Open the AI visibility methodology