Entity and Brand Mention Consistency Audit for AI Visibility
Audit how your brand, products, and people are described across owned and third-party sources, and find inconsistencies that could confuse AI answers. Returns an evidence table, fix list, and verification plan.
You getStatus and assumptions, discovery plan table and consistency table
Role: You are an entity SEO and AI visibility specialist who audits how a brand is represented across the web.
Context: Alder & Finch Legal Software (example value for {brand}) (alderfinch.example.com (example value for {brand_url})) offers contract management software for in-house legal teams (example value for {offering}). Answer engines sometimes describe it inaccurately or omit it. The user decision: which inconsistencies to fix first and on which properties.
Task: Audit entity consistency using the sources I provide or the searches I run. If live browsing is not enabled in this Gemini session, state that and provide a discovery plan only until I paste results.
Inputs:
- Brand and URL: Alder & Finch Legal Software (example value for {brand}), alderfinch.example.com (example value for {brand_url})
- Offering and category: contract management software for in-house legal teams (example value for {offering})
- Canonical facts (founding year, HQ, category, product names, leadership, pricing model): Founded 2016; HQ Chicago; products: Alder Contracts and Finch Review; pricing per seat, annual (example value for {canonical_facts})
- Third-party properties to check: LinkedIn, Crunchbase, G2, Capterra, Wikipedia or Wikidata if present (example value for {properties})
- Pasted evidence: page excerpts, AI answers about the brand, profile text, structured data snippets, with dates checked.
Workflow:
1. Declare browsing status; list assumptions separately.
2. Discovery plan: searches to run, such as 'Alder & Finch Legal Software (example value for {brand})' alone, 'Alder & Finch Legal Software (example value for {brand}) reviews', 'Alder & Finch Legal Software (example value for {brand}) vs [competitor]', 'Alder & Finch Legal Software (example value for {brand}) founder', 'site:alderfinch.example.com (example value for {brand_url}) about', plus checks of each property in LinkedIn, Crunchbase, G2, Capterra, Wikipedia or Wikidata if present (example value for {properties}). Add a prompt set to ask Gemini and one other answer engine, 'What is Alder & Finch Legal Software (example value for {brand})?' and 'Who is Alder & Finch Legal Software (example value for {brand}) best for?', and record the answers verbatim with dates.
3. Compare each observation to Founded 2016; HQ Chicago; products: Alder Contracts and Finch Review; pricing per seat, annual (example value for {canonical_facts}). Mark Match, Mismatch, Missing, or Unverifiable.
4. Identify root-cause hypotheses (outdated profiles, conflicting product names, thin About page, missing Organization schema, ambiguous naming with other entities).
5. Separate Evidence, Inference, Contradictions, and Unanswered questions.
6. Prioritize fixes by likely visibility impact and effort.
Constraints:
- Use only provided canonical facts; do not guess missing facts.
- Cite source title, publisher, URL, and date checked, or mark missing.
- Do not assume schema markup is present or valid without a pasted snippet or test result.
- Do not claim a fix will change AI answers; propose it as a testable hypothesis.
Output format:
1. Status and assumptions
2. Discovery plan table: Check | Search or URL | What to record
3. Consistency table: Fact | Canonical value | Source title | Publisher | URL | Observed value | Status | Date checked
4. AI answer log: Engine | Prompt | Answer summary | Errors or omissions | Date
5. Root-cause hypotheses with confidence
6. Fix list: Priority | Property | Change | Owner | Verification method
7. Evidence / Inference / Contradictions / Unanswered questions
8. 30-day re-test plan
Acceptance criteria:
- Every canonical fact is checked on at least the owned site and one third-party property or is marked Unverifiable.
- Each fix names the exact property and field to change.
- The re-test plan lists the same prompts to repeat for comparison.
Quality checks:
- No invented facts about the brand.
- Mismatches are quoted from the source, not paraphrased vaguely.
- Hypotheses are distinguished from findings.
- Risks (name ambiguity, conflicting legacy sources) are noted.
Dashed parts are example values, so the prompt runs as copied. Change them in Fill in the blanks.
Prepare a one-page canonical facts sheet first; the audit is only as good as that baseline. Check structured data with a dedicated testing tool and paste the results, rather than asking Gemini to infer markup.
Design a repeatable panel of prompts, engines, and scoring rules to measure brand mentions and citations in AI answers. Produces a baseline protocol, a scoring sheet, and a decision rule for reporting changes.
Create a content brief for one target query using the sources AI answers and search results actually rely on. Returns the questions to answer, evidence to cite, structure, and acceptance checks for the writer.
Map which domains and source types answer engines rely on for a topic cluster, so you can decide where to publish, earn mentions, or build original research. Includes search steps, a citation table, and confidence labels.
Find which sources AI answers cite for your priority queries, where your brand is missing, and which page fixes to fund first. Built for Gemini with source-discovery steps and a labeled evidence table.