The five metrics, defined
Over twelve weeks we worked with Shopify stores doing between $80K and $4M per month across fashion, beauty, homewares, and supplements. We ran a standardised battery of 200 buyer-intent prompts weekly across ChatGPT, Gemini, Perplexity, Claude, and Microsoft Copilot, then mapped every signal back to Shopify revenue. Five metrics consistently predicted AI-driven revenue. Everything else was noise.
METRIC 01
Citation Rate
Citation rate is the percentage of prompts in your tracked set where your brand or a product URL appears in the AI-generated answer. It is the headline metric on any serious AI-visibility dashboard — the equivalent of keyword ranking in traditional SEO.
Define a prompt set of at least 50 buyer-language queries relevant to your category. Run them weekly across every major AI platform. If you appear in 8 of 50 answers your citation rate is 16%. Track it by platform, by product category, and against your prior 30-day baseline.
Why it matters: A citation rate below 10% on a well-defined prompt set means AI models lack enough structured, authoritative information about your products to recommend them confidently. The fix starts with data — JSON-LD schema, an accurate llms.txt, and server-side rendered product content — not with ad spend.
METRIC 02
Share of Voice (AI)
AI share of voice measures your brand’s citation count as a percentage of total citations across you and your top five competitors, for the same prompt set. A citation rate of 16% sounds decent until you discover that your nearest competitor is at 54%.
Track share of voice by prompt category — ingredient queries, price-point queries, use-case queries, and brand-comparison queries often tell very different stories. A brand can dominate ingredient prompts and be invisible in comparison prompts, which is exactly the gap a competitor’s content team will exploit.
Why it matters: AI answers rarely list more than three brands. If your share of voice is below 20% in your category you are competing for one of the final slots, and small content and schema improvements can move you from invisible to consistently cited within weeks. Tools like Ryze AI’s visibility dashboard track competitive share of voice automatically across all five major AI platforms.
METRIC 03
AI Sentiment Score
Being mentioned is not enough. An AI answer that says “Brand X has mixed reviews on ingredient quality” while recommending a competitor is worse than not being cited at all. Sentiment tracks the qualitative tone of every mention: positive, neutral, or negative, broken down by prompt category.
Score sentiment on a simple three-tier scale. Positive means the AI recommended you, described your product favourably, or cited you as a top pick. Neutral means you were mentioned without a recommendation. Negative means you appeared in a comparison and lost, or were cited as a cautionary example.
Why it matters: Sentiment is almost always a content and authority problem. AI models synthesise what they have been trained on and what they can currently retrieve. Negative or neutral sentiment usually traces to thin product descriptions, absent review schema, or competitor content that makes stronger factual claims. Sentiment trending downward across platforms is an early warning sign weeks before revenue dips.
METRIC 04
AI Click-Through Rate (AI-CTR)
Citation rate tells you how often you appear. AI-CTR tells you how often appearing translates into a visit. This is measured via referrer analysis in GA4 — sessions arriving from chatgpt.com, perplexity.ai, gemini.google.com, claude.ai, and bing.com/chat — divided by your estimated citation volume for the same period.
Google Search Console’s new Generative AI section (Performance > Search Results > Generative AI) now surfaces impressions for AI Overviews and AI Mode, giving you a native starting point for click-through calculation on Google’s AI surfaces. Combine that with direct referrer data from the other platforms to build a complete picture.
Why it matters: A high citation rate with low AI-CTR means you are being mentioned but not linked, or your product URLs are being paraphrased rather than surfaced. The fix is structural: ensure every product page has a canonical URL clearly associated with the brand name, and that your llms.txt and sitemap give AI crawlers the right link to surface. See our guide on tracking AI visibility for Shopify beauty brands for a worked example.
METRIC 05
AI-Attributed Revenue (Sessions, CVR, AOV, Contribution Margin)
The five metrics that belong on a Shopify AI-visibility dashboard culminate here: revenue that can be directly attributed to AI-referred sessions. This is the finance view that makes the channel real to a CFO. Track four sub-metrics as a bundle: AI-attributed sessions (volume), conversion rate of those sessions, average order value, and contribution margin per AI-referred order.
AI-referred shoppers convert at 14.2% on average versus 2.8% for Google organic, according to 2026 benchmarks from The Hope Factory. They also tend to arrive with higher purchase intent — they have already been pre-sold by the AI answer — which frequently drives AOV 15–25% above site average on stores we audited.
Why it matters: Without revenue attribution the first four metrics are interesting but not fundable. Attribution closes the loop: you can now calculate a cost-per-AI-cited-session, a channel ROAS equivalent, and justify investment in schema, content, and structured data improvements the same way you justify paid media spend. Ryze AI builds this attribution layer automatically inside your Shopify analytics, connecting bot-log data to order revenue without manual tagging.
HOW WE SCORED THE TOOLS BELOW
We evaluated 10 approaches against the same five metrics framework above. Scoring criteria:
- Prompt coverage breadth — how many query types and AI platforms does it monitor automatically?
- Metric completeness — does it cover all five metrics or just citation rate?
- Actionability — does it fix visibility gaps, or just report them?
- Shopify-native integration — catalogue import, revenue attribution, automated fixes?
- Measurable AI referral lift against each store’s prior 90-day AI traffic baseline
No vendor paid for placement. Ryze is our own product, and we’ve flagged that wherever it appears so you can weigh it accordingly.