Skip to main content

Direct answer

Start with the decisions the report should support. Select a small set of product metrics, traffic observations, and qualitative findings; document every filter and denominator; then use Exports, Looker Studio, or Shared Links to present a reproducible view.

One-page executive report

AI visibility in France · Fictional example, 2026 Decision requested. Approve one guide to switching providers. Five of the eight prompts that still received no mention of our brand asked about migration. Scope. We tracked the same 20 French-language comparison prompts, none naming a brand, once a week on ChatGPT (Core) and Perplexity (Core) in France. Each four-week period produced 160 answers. The list of direct competitors stayed fixed. Coverage counts distinct prompts with at least one answer mentioning our brand out of all answered prompts. Mention Rate counts answers mentioning us out of those mentioning us or a direct competitor, 120 in each period. Changes are in percentage points. What changed. Our brand appeared on four more prompts. Across eligible answers, it appeared 35% of the time. These figures measure visibility. We did not collect traffic or sales data. Next review. The content lead will publish the migration guide by November 6. The analytics lead will repeat both measures for November 9 to December 6 and check which of the five migration prompts mention us. The example uses Overview and Response Analysis.

Data required

Use a fixed reporting period, stable prompt cohort, providers and models, topics or tags, SELF/DIRECT classifications, stored answers and citations, any connected traffic data, and the previous comparable period. Record the source and formula for every metric.

Workflow

1

Define the audience and decision

State whether the report supports brand monitoring, content prioritization, technical work, or commercial measurement. Exclude metrics that do not inform that decision.
2

Freeze the cohort

Record prompts, providers, models, topics, geography, and period. Note changes from the comparison period.
3

Build the evidence set

Combine product metrics with representative answers and separate traffic observations. Use Exports or documented Looker Studio sources.
4

Write the narrative

Report what changed, where it changed, material limitations, and a short list of owners and testable next actions.

Interpretation

Product metrics must retain their exact numerator, denominator, filters, and unit. Mentions, citations, Share of Voice, Coverage, sentiment, crawler occurrences, and referral sessions answer different questions. A change in one should not be described as causing a change in another.

Possible next actions

  • If one metric moves sharply, test whether the prompt, provider, model, or competitor mix changed before assigning an owner.
  • If an executive question lacks a valid measure, label it as an open question instead of substituting a proxy.
  • If recurring reporting is needed, test a Looker Studio or API workflow with documented field mappings.

Limitations

Qwairy observes configured prompts and stored provider outputs. Results are not exhaustive and can change with models, sources, and monitoring configuration. The REST API is a separate contract; verify endpoint fields and formulas before combining API and product data.

Looker Studio setup

Connect documented data sources for reporting.

REST API

Review API authentication, resources, and contracts.

Shared Links

Share a filtered product view.

KPIs for GEO

Choose metrics with explicit definitions.

ROI comparison

Build an assumption-led commercial view.