AI Visibility

Erlin runs a set of prompts on your behalf across six AI platforms (ChatGPT, Perplexity, Gemini, Claude, and Google’s AI Overview and AI Mode), captures every response, and measures five things: whether your brand was mentioned, where it was ranked, what share of the list was yours, whether your site was cited, and what tone AI used. The results appear on the AI Visibility dashboard as five metric cards, trend charts, a platform breakdown, and a competitor leaderboard.

​

Platforms Erlin tracks

Erlin runs your prompts on up to six platforms, split between AI chat models and Google search features. Chat models answer the prompt directly:

• ChatGPT (OpenAI)
• Perplexity
• Gemini (Google)
• Claude (Anthropic)

Google search features show how your brand appears in AI-powered Google results:

• AI Overview: the AI-generated summary Google places at the top of a search results page. Google does not trigger it for every query, so some prompts return no AI Overview result.
• AI Mode: Google’s conversational AI search. It always returns an answer, with its own sources and follow-up questions.

Which platforms you can track depends on your plan: Starter covers ChatGPT and Perplexity, Standard adds Gemini, and Pro and Enterprise add Claude, AI Overview, and AI Mode. Higher BYOK tiers include AI Overview and AI Mode as well. See Plans and usage for what your plan includes.

​

What Erlin captures per response

For every prompt Erlin runs on each platform, it records:

• Whether your brand was mentioned in the response
• The position your brand was given in any recommendation list
• What share of the brands listed in the response are yours
• Whether the response cited a URL on your domain
• The tone of any language used about your brand

Erlin runs each prompt across every platform your plan enables. Tracking 30 prompts on four platforms gives you 120 responses per scan. Each response feeds one data point into each of the five metrics: Visibility Score, Citation Rate, Share of Voice, Average Position, and Sentiment.

​

How a scan works

A scan is one full run of every tracked prompt on every enabled platform. When you trigger a scan, Erlin:

1. Sends each prompt to each AI platform
2. Parses the full response to extract product mentions, brand names, and source URLs
3. Classifies each cited source by type: article, aggregator, news, forum, video, or reference
4. Scores the tone of any mention of your brand using LLM-based analysis
5. Stores the result as one tracked response
6. Recalculates all dashboard metrics once the last prompt completes

After the scan finishes, Erlin triggers the citation analysis pipeline automatically. That pipeline populates the Citations module with source-level data: which URLs AI cited, whether your brand appears on them, and how influential each source is. Start a scan manually from the dashboard header, or set Erlin to run scans on a schedule via Prompt tracking.

​

Metric cards

The top of the dashboard shows five cards, one per metric. Each card displays the current value, a green or red direction indicator, and the change since the previous period. Each metric has a dedicated page with the formula, worked example, score ranges, and actions to improve it.

​

Trend charts

Below the metric cards, Erlin charts how each metric has moved over the selected time range. Each line uses a distinct color matching its card. Hover any data point to see the exact value, the number of responses in that period, and the date it represents. A rising line means that metric is improving week over week. A falling line means you are losing ground. Trend charts are the fastest way to spot a change and investigate before the gap widens.

​

Platform Breakdown

The Platform Breakdown table shows your Visibility Score, citation rate, and average position per platform side by side. Use it to compare every platform your plan tracks at a glance: ChatGPT, Perplexity, Gemini, Claude, AI Overview, and AI Mode. Perplexity almost always shows citations because it displays sources with every answer, and Google’s AI Overview and AI Mode return their own source links too. ChatGPT and Claude sometimes answer without citing sources, so citation rate on those platforms may be lower even when they mention your brand. A platform where Visibility Score is high but citation rate is low means AI talks about you there but pulls its source links from other content. That points to specific content work on that platform.

​

Competitor Leaderboard

The Competitor Leaderboard ranks every brand AI mentioned alongside yours. It has two tabs: Top Competitors shows every brand that appeared in AI responses to your tracked prompts, ranked by share of voice. Brands you did not know you competed with show up here. If a brand you have never heard of has 25% share of voice on your prompts, AI considers them relevant to your audience. Your Competitors shows only the brands you have added in brand settings. Use this tab for direct, intentional benchmarking against the specific rivals your team tracks. Each row shows average position, Visibility Score, and share of voice for the selected time range. Columns are sortable. Your own brand appears in the table flagged with a “YOU” badge.

​

Time range

The time-range selector in the dashboard header controls the entire page. Options: last 7, 14, 30, or 90 days. When you change the range, every card, chart, and table recalculates. Use 7 days to see immediate impact after publishing new content. Use 30 or 90 days to see whether a competitor is gaining or losing ground over time.

​

Filters

Beyond time range, the dashboard supports three filters that apply simultaneously:

• Platform: narrow to one platform (useful when you want to isolate one platform’s data)
• Folder: focus on a specific prompt group (one product line, one funnel stage, one competitor set)
• Competitor: overlay a competitor’s trend lines on your charts for direct comparison

All filters apply to every chart and table at once. Metric cards, trends, platform breakdown, and competitor leaderboard always stay in sync.

​

How to refresh data

Three ways to bring in fresh data: Run a scan. The Run button in the dashboard header executes every tracked prompt immediately. Use this when you have published new content and want to measure the impact. Set a schedule. The Schedule dropdown in Prompt tracking runs scans on a recurring interval: every 7, 14, or 30 days. Once enabled, Erlin runs scans without manual intervention. Refresh the cache. Erlin caches dashboard responses for speed. The refresh icon in the header forces a re-read from the latest stored data without triggering a new scan. Use this when a scan recently completed and the numbers have not updated yet.

​

Related