How we measure AI Visibility (and why it's about to matter more than rank)

When a B2B buyer wanted to find software in 2014, they Googled. When they wanted to find software in 2020, they still Googled — and then read a G2 review or a comparison post. When they want to find software in 2026, they increasingly just ask Claude, ChatGPT, Perplexity, or Gemini, "what's the best X for Y?" — and they take whatever the LLM names.

If your brand isn't in that answer, you're invisible. And the LLM doesn't show the buyer what it could have said.

Why this is different from rank tracking

Traditional SEO tools track your position on Google for keywords. That's a known job, well-served by Ahrefs, Semrush, and a dozen others. But it's downstream of the new reality: the LLM's answer is now the first answer the buyer sees, and the citations the LLM picks are an entirely different distribution than Google's SERP.

Two reasons LLMs cite differently from Google ranks:

1. Training-data recency. Most LLMs are trained on a snapshot 6-18 months old. Pages that ranked when the snapshot was taken get cited even after they've fallen on Google. New pages that rank well today may not show up in LLM answers for months.
2. Real-time grounding tools. Perplexity, Gemini, and ChatGPT (when web search is enabled) ground answers in live search results — but they synthesize across several sources rather than reproducing rank order. A page ranked #5 with great structured data + good entity signals can get cited over a page ranked #1 with weak signals.

How Slope probes

Slope's AI Visibility feature builds a per-project prompt set — typically 15-30 prompts seeded from your brand, competitors, and your top GSC queries. You curate the prompts so they match what your buyers actually ask. Then on a daily schedule (or on-demand), Slope sends each prompt to every configured provider:

• Perplexity API — citations are returned natively (the most usable signal).
• Gemini API with tools: [{ google_search: {} }] — citations come from groundingMetadata.groundingChunks.
• Anthropic Claude API — Claude doesn't ground natively, so we extract URLs from the answer text with a regex (best-effort citations).
• ChatGPT, Google AI Overviews, Bing Copilot — coming via SerpAPI integration.

What we extract from each answer

• Brand mention. Is your brand name in the answer text? At what character position (proxy for prominence)?
• Brand citation. Is your domain in the citation list? At what position (1st cited, 3rd cited, etc.)?
• Competitor mentions. For each of your tracked competitors, did they get name-dropped or cited? Where?
• Light sentiment. The 200 chars around your brand — are there negative-signal words ("bad", "avoid", "disappointing") or positive ("best", "leading", "recommended")?

The gap → fix loop

Prompts where your brand is missing but competitors are present become R7 opportunities — theai_visibility rule family. Each one suggests a specific fix: "create a comparison page for X", "publish a guide on Y", "add structured data + author bio to existing page Z".

These opportunities push to GitHub like any other Slope opportunity. The dev team (or your AI coding agent) ships the page. The next weekly probe re-tests the prompt. If your brand now appears, the opportunity gets marked completed and the impact worker records the visibility delta.

What we're still figuring out

Honest list of open questions:

• Causation vs correlation. A page that gets cited by Perplexity also tends to rank on Google — separating the contributions is hard.
• Provider stability. Gemini's grounding API has changed twice in the past year. We have to rebuild the citation parser when it does.
• Cost economics. Probing 25 prompts × 4 providers × daily is ~3,000 API calls/month per project. At Perplexity Sonar + Gemini Flash + Claude Sonnet pricing, that's ~$2-5/month/project — fine on Growth tier, but volume on Agency tier needs careful caching.

Try it

AI Visibility is included on Slope's Growth tier ($149/mo). You bring your own LLM keys (Anthropic + OpenAI + Perplexity + Gemini), so the LLM cost is yours and Slope's subscription price stays predictable.

Start free and upgrade once you want the AI Visibility loop running on your prompts.