01
Scope
Atlas measures how often selected AI assistants name businesses in response to a versioned prompt set for a city and category. The panel is fixed within an edition and may change only in a newly versioned edition. Atlas does not measure service quality, market share, search traffic, or buyer conversion.
- Observation
- One prompt × assistant × repetition response
- Scan run
- One dated measurement of the full panel
- Snapshot
- One published ranking derived from a validated run
02
Sampling
Prompts describe buyer situations, constraints, research questions, and selection decisions. Each prompt is sent independently to every assistant in scope and repeated according to the prompt-set protocol.
Crisis
Immediate event
Advice
Whether to act
Situation
Specific circumstances
Constraints
Budget or access
Learning
Process question
Decision
Who to consider
Validation
Check a named firm
Follow-up
Later conversation turn
Bucket names and counts are versioned. Public pages show panel structure and aggregate results; limited examples may appear in dated research reports. Complete prompt text and response-level records are available only through paid audits or licensed research access.
03
Ranking inclusion
Ranking layer
Prompts that request a business recommendation or evaluation. Qualifying business mentions in these responses determine rank and response mention rate.
Editorial layer
Advice, crisis, learning, or constraint prompts that do not request names. Samples may be published, but they are excluded from ranking calculations.
04
Entity extraction
- 1Parse business and professional names from each response.
- 2Match names to Atlas records using normalized names, aliases, domains, and canonical entity links.
- 3Classify the entity and its role in that answer using an immutable, reviewed dataset and ranking-policy version.
- 4Rank businesses a buyer could hire. Show directories, associations, comparison sites, and other sources in a separate list.
- 5Hold unresolved entity matches out of public business rankings until they can be reviewed.
Response-level evidence is retained for auditability. Public pages summarize that evidence: businesses appear in the ranking, while directories, associations, and other research sources appear in a separate Sources section.
05
Metrics
Mentions
Eligible responses that named the canonical business. One business counts at most once per response.
Response mention rate
mentions ÷ eligible responses. Reported with a 95% Wilson confidence interval.
Mention share
business mentions ÷ all business mentions in the snapshot.
Assistant coverage
Number of tested assistants that named the business at least once.
Source mentions
How often AI answers cite a directory, association, comparison site, or other research source. Source mentions are reported separately from business rankings.
Rank orders businesses within one snapshot. Comparisons across categories require the underlying rates and sample sizes.
06
Models and snapshots
Claude
anthropic/claude-sonnet-4.6
ChatGPT
openai/gpt-4o
Gemini
google/gemini-2.5-flash
Perplexity
perplexity/sonar
Snapshots retain their prompt-set version, model set, source runs, sample size, publication date, entity-classification dataset, ranking policy, builder version, and methodology hash. Published snapshots are not rewritten. A model, prompt-set, classification, or ranking-policy change creates a new immutable edition.
Model change · August 3, 2026
Gemini changed from gemini-2.5-pro to gemini-2.5-flash; Perplexity changed from sonar-pro to sonar. Results on opposite sides of this change are not directly comparable.
07
Prompt sets
Active sets are shown with their published bucket counts. Expand a set for details.
divorce-attorneys v3.1 (8 buckets, 141 prompts)8 buckets · 141 questions v3.1⌄
Advice
12 questions
Crisis
12 questions
Decision
33 questions
Learning
10 questions
Follow Up
16 questions
Situation
23 questions
Validation
25 questions
Constraints
10 questions
personal-injury v3.1 (8 buckets, 155 prompts)8 buckets · 155 questions v3.1⌄
Advice
12 questions
Crisis
16 questions
Decision
33 questions
Learning
10 questions
Follow Up
19 questions
Situation
27 questions
Validation
25 questions
Constraints
13 questions
immigration v3.1 (8 buckets, 155 prompts)8 buckets · 155 questions v3.1⌄
Advice
11 questions
Crisis
15 questions
Decision
36 questions
Learning
11 questions
Follow Up
17 questions
Situation
26 questions
Validation
24 questions
Constraints
15 questions
employment-plaintiff v3.1 (8 buckets, 147 prompts)8 buckets · 147 questions v3.1⌄
Advice
12 questions
Crisis
12 questions
Decision
35 questions
Learning
11 questions
Follow Up
17 questions
Situation
24 questions
Validation
24 questions
Constraints
12 questions
cosmetic-dentistry v1.0 (8 buckets)8 buckets · 141 questions v1.0⌄
Advice
12 questions
Crisis
12 questions
Decision
33 questions
Learning
10 questions
Follow Up
16 questions
Situation
23 questions
Validation
25 questions
Constraints
10 questions
cosmetic-surgery v1.0 (8 buckets)8 buckets · 141 questions v1.0⌄
Advice
12 questions
Crisis
12 questions
Decision
33 questions
Learning
10 questions
Follow Up
16 questions
Situation
23 questions
Validation
25 questions
Constraints
10 questions
ivf v1.0 (8 buckets)8 buckets · 141 questions v1.0⌄
Advice
12 questions
Crisis
12 questions
Decision
33 questions
Learning
10 questions
Follow Up
16 questions
Situation
23 questions
Validation
25 questions
Constraints
10 questions
estate-planning v1.0 (8 buckets)8 buckets · 141 questions v1.0⌄
Advice
12 questions
Crisis
12 questions
Decision
33 questions
Learning
10 questions
Follow Up
16 questions
Situation
23 questions
Validation
25 questions
Constraints
10 questions
real-estate-law v2.0 (5 buckets)5 buckets · 25 questions v2.0⌄
Process
5 questions
Specific Need
5 questions
Conversational
5 questions
Style Approach
5 questions
Broad Authority
5 questions
wealth-management v1.0 (8 buckets)8 buckets · 141 questions v1.0⌄
Advice
12 questions
Crisis
12 questions
Decision
33 questions
Learning
10 questions
Follow Up
16 questions
Situation
23 questions
Validation
25 questions
Constraints
10 questions
advisor-discovery v1.0 (8 buckets)8 buckets · 70 questions v1.0⌄
Advice
7 questions
Crisis
5 questions
Decision
18 questions
Learning
7 questions
Follow Up
6 questions
Situation
10 questions
Validation
12 questions
Constraints
5 questions
cpa v1.0 (8 buckets)8 buckets · 141 questions v1.0⌄
Advice
12 questions
Crisis
12 questions
Decision
33 questions
Learning
10 questions
Follow Up
16 questions
Situation
23 questions
Validation
25 questions
Constraints
10 questions
Archived versions (4)
- Legacy v1.0 (divorce-attorneys) · v1.0 10 questions · Jun 25, 2026
- divorce-attorneys v2.0 (5 buckets) · v2.0 25 questions · Jun 25, 2026
- divorce-attorneys v3.0 (5 buckets, 100 prompts) · v3.0 100 questions · Jun 25, 2026
- Legacy legal/real-estate-law · v1.0 Legacy prompt text not retained · Jul 23, 2026
08
Access and corrections
Public
- Rankings and confidence intervals
- Per-assistant counts
- Prompt-set versions
- Professional mention tables
Paid audit or licensed research access
- Response-level records
- Per-prompt breakdowns
- Historical exports
- Custom scan scopes
For data access, factual corrections, or removal requests, email support@viclaro.app. Factual corrections include names, websites, entity matches, and category assignments. Ranking positions are not changed on request.
Atlas and per-brand audits
Atlas reports category-level results from versioned prompt sets. A per-brand audit runs ChatGPT, Claude, Gemini, and Perplexity against a narrower deck of buyer prompts scoped to one brand. It reports the observed prompt-level mention rate alongside a page-side content-readiness estimate: a model-scored heuristic of whether the audited page contains clear, relevant answer material. That estimate is not a probability or an upper limit on future citations. A gap between the measures may reflect off-page authority, retrieval behavior, category positioning, brand familiarity, or sample variance. Draft answer blocks for uncited prompts are editorial starting points and require factual and compliance review before publication. Successive audits use the same prompt deck and model definitions when possible, but changes can still reflect model, retrieval, and sampling variance; movement between audits should not be attributed to shipped content alone.