By Satish K, Co-Founder and CEO, Astiva AI · Updated April 2026
Write 10 real buyer-intent queries. Use the shape that buyers actually type: "best X for Y", "X vs Y", "how to do X with limited budget". Skip branded queries. The unbranded, category-level queries are the ones where the AI gets to choose whom to recommend.
Run each query across ChatGPT and Perplexity. Start with these two because they drive the majority of AI-generated brand discovery today. Repeat each query 2-3 times to capture response variance. Copy each full answer verbatim into a spreadsheet.
Expand to Claude, Gemini, Grok, Meta AI, and Mistral AI. Different AI surfaces use different training cutoffs and retrieval strategies. A brand recommended on ChatGPT may be invisible on Claude if the cited sources do not appear there.
Log brand position, sentiment, and competitors. For each response, record: was the brand named at all (Visibility %), at what ordinal position (Average Position), with what tone (Sentiment), and which competitors were named alongside or instead.
Diagnose each miss. Misses fall into three buckets: (a) citation gap; (b) positioning gap; (c) competitor displacement. Each bucket needs a different fix.
Fix the gaps with citation-ready content. Ship structured, source-citable content. Perplexity reflects new citations within days; ChatGPT and Claude surface new content more gradually. Re-run queries after 7 to 14 days for Perplexity, and after 30 days for the full platform picture.
Why this matters
AI search collapses 10 blue links into one recommendation. If your brand is not named, you are not in the consideration set regardless of your Google rank. Teams that audit AI visibility on a recurring cadence catch new competitors, model updates, and sentiment drift before they show up in pipeline numbers.
Common questions
How often should I re-run the audit?
Quarterly at minimum for stable brands; monthly if you are in an active category where a model update or competitor launch can shift the naming moment. Astiva runs the same prompts daily automatically, so you get continuous monitoring rather than snapshot audits. Daily tracking catches model update impacts and competitor gains within 24 hours rather than weeks later.
What counts as a "good" Visibility %?
It depends on category density and competitive set size. Highly competitive SaaS categories land leaders at 40-60% visibility; niche verticals can hit 80%+ for the dominant brand. Category average is a more meaningful benchmark than an absolute number. Astiva's methodology page documents reporting windows and provides category benchmark context for the most common verticals.
Do I need a paid tool to do this?
For a one-time audit across 2-3 platforms with 10 manual queries, no paid tool is required. For ongoing, daily tracking across all 10 AI surfaces with automated competitor benchmarking, Share of Voice calculation, and GA4 revenue attribution, a purpose-built tool like Astiva saves significant time each week and catches shifts you would miss in a quarterly manual audit.