KPIs
Sitefire tracks your AI visibility across four surfaces: the answers AI models give, the citations they pull from, the bots that crawl your site, and the traffic those answers send back to you. This page is the canonical reference for every metric you’ll see across the product.
At a glance
| Metric | Unit | What it measures |
|---|---|---|
| Visibility Score | % | Share of AI answers mentioning your brand |
| Share of Voice | % | Your mentions vs. all brand mentions |
| Average Position | Rank (1 = best) | Average rank when you’re mentioned |
| Citation Rate | % | Share of answers citing your pages |
| Citation Share | % | Your citations vs. all citations |
| Bot Requests | Count | AI bot visits to your site |
| Active Bots | Count | Distinct AI bots crawling you |
| Top Provider | Name | Provider sending the most crawl traffic |
| Top Page | Path | Most-crawled page |
| AI Sessions | Count | GA4 visits from AI engines |
| Conversions | Count | GA4 conversions from AI traffic |
| Conversion Rate | % | AI-referred sessions that convert |
| Engagement Rate | % | Engaged AI-referred sessions |
| Bounce Rate | % | Non-engaged AI-referred sessions |
| Avg. Duration | Seconds | Average AI-referred session length |
Visibility metrics
Visibility Score
Unit: percentage (0-100%)
The share of AI answers that mention your brand, weighted by how much each prompt matters to you.
Why it matters - Tells you how often your brand shows up in AI answers at all. The headline metric, before you worry about how prominently.
Formula
Score = Σ (prompt_weight × mention_rate) / Σ prompt_weight
prompt_weight = prompt_volume × model_weight
prompt_volume = topic_volume / active_prompts_in_topicInputs
| Input | Description |
|---|---|
topic_volume | Monthly search volume for the topic. Topics are the broader subjects you track; each topic contains one or more prompts. |
active_prompts_in_topic | Number of active prompts under the topic. The topic’s volume is split evenly across them, so adding more prompts to a topic dilutes each prompt’s weight. |
model_weight | Your configured model mix. Weights across ChatGPT, Perplexity, Gemini, and other supported models, summing to 100%. |
mention_rate | The share of answers on that prompt/model where your brand is mentioned (0-100%). |
Example
You’re Northwind, a CRM for small businesses. You track Contoso and Fabrikam as competitors. You’ve set up two topics, each with a single active prompt:
| Topic | Topic volume | Active prompts | Prompt volume | Prompt text |
|---|---|---|---|---|
| Small-business CRM tools | 1,000 | 1 | 1,000 ÷ 1 = 1,000 | A: “best CRM for small business” |
| Sales pipeline tracking | 500 | 1 | 500 ÷ 1 = 500 | B: “how to track sales pipeline” |
Each topic’s volume splits evenly across its active prompts. In this example each topic has only one, so each prompt gets the full topic volume. If a topic had 3 active prompts, each would get a third of the topic volume.
Model weights: ChatGPT 80%, Perplexity 20%. Each prompt is asked on each model, so you get 4 answers.
Answer 1 — Prompt A on ChatGPT
For small businesses, these CRMs come up most often:
- Northwind — strong pipeline management with a generous free tier [1]
- Contoso — simple contact management, low learning curve [2]
- Fabrikam — broader feature set, geared toward growing teams
Sources: [1] northwind.com/features · [2] contoso.com
Answer 2 — Prompt A on Perplexity
The most-recommended CRMs for small businesses:
- Contoso — affordable, used by thousands of small teams [1]
- Northwind — richer feature set for active pipeline tracking [2]
Sources: [1] contoso.com · [2] northwind.com/features · [3] fabrikam.com
Answer 3 — Prompt B on ChatGPT
Core practices for tracking a sales pipeline:
- Contoso’s guide recommends a weighted-probability approach [1]
- Fabrikam’s playbook emphasizes stage-by-stage forecasting [2]
Sources: [1] contoso.com/guide · [2] fabrikam.com
Answer 4 — Prompt B on Perplexity
Three tools come up for pipeline tracking:
- Northwind — visual kanban with stage probabilities [1]
- Contoso — solid CRM with a built-in pipeline view
- Fabrikam — enterprise-grade forecasting
Sources: [1] northwind.com/tracking
Extracted, that’s:
| # | Prompt | Model | Mentions (in order) | Citations |
|---|---|---|---|---|
| 1 | A | ChatGPT | Northwind, Contoso, Fabrikam | northwind.com/features, contoso.com |
| 2 | A | Perplexity | Contoso, Northwind | contoso.com, northwind.com/features, fabrikam.com |
| 3 | B | ChatGPT | Contoso, Fabrikam | contoso.com/guide, fabrikam.com |
| 4 | B | Perplexity | Northwind, Contoso, Fabrikam | northwind.com/tracking |
Northwind’s mention rate per prompt:
- Prompt A: mentioned on ChatGPT and Perplexity → 80% + 20% = 100%
- Prompt B: mentioned only on Perplexity → 0% + 20% = 20%
Score = (1,000 × 100% + 500 × 20%) / 1,500 = 73.3%Related: Share of Voice · Average Position
Share of Voice
Unit: percentage (0-100%)
Your mentions as a percentage of all brand mentions in the same AI answers.
Why it matters - Tells you whether you’re gaining ground on competitors. Visibility Score can rise across the whole category at once; Share of Voice only rises when you’re winning a bigger slice of mentions than competitors are.
Formula
Share of Voice = your_mentions / total_mentionsInputs
| Input | Description |
|---|---|
your_mentions | Count of answers in the selected prompts and date range that mention your brand. |
total_mentions | Count of mentions of any tracked brand in the same answers. |
Example
You’re Northwind, a small-business CRM. You track Contoso and Fabrikam as competitors. Across two prompts asked on ChatGPT and Perplexity, you collect 4 answers:
| # | Prompt | Model | Mentions (in order) | Citations |
|---|---|---|---|---|
| 1 | A | ChatGPT | Northwind, Contoso, Fabrikam | northwind.com/features, contoso.com |
| 2 | A | Perplexity | Contoso, Northwind | contoso.com, northwind.com/features, fabrikam.com |
| 3 | B | ChatGPT | Contoso, Fabrikam | contoso.com/guide, fabrikam.com |
| 4 | B | Perplexity | Northwind, Contoso, Fabrikam | northwind.com/tracking |
Count mentions:
- Total mentions across all answers: 3 + 2 + 2 + 3 = 10
- Northwind mentions: 1 + 1 + 0 + 1 = 3
Share of Voice = 3 / 10 = 30%Northwind’s Visibility Score on this data is 73.3% but Share of Voice is 30%. Northwind shows up in most answers, but is only one of several brands named each time.
Related: Visibility Score · Average Position
Average Position
Unit: rank (1 is best; lower is better)
When your brand is mentioned, the average rank it appears at within the AI answer.
Why it matters - Being mentioned 7th in a list of 10 is very different from being the first recommendation. Shows how prominently you appear, not just whether you appear.
Formula
Average Position = mean(rank) across all answers where your brand is mentionedExample
You’re Northwind, tracking Contoso and Fabrikam. Across two prompts asked on ChatGPT and Perplexity, you collect 4 answers:
| # | Prompt | Model | Mentions (in order) |
|---|---|---|---|
| 1 | A | ChatGPT | Northwind, Contoso, Fabrikam |
| 2 | A | Perplexity | Contoso, Northwind |
| 3 | B | ChatGPT | Contoso, Fabrikam |
| 4 | B | Perplexity | Northwind, Contoso, Fabrikam |
Northwind’s rank in each answer where it’s mentioned:
- Answer 1 - rank 1
- Answer 2 - rank 2
- Answer 3 - not mentioned, skipped
- Answer 4 - rank 1
Average Position = (1 + 2 + 1) / 3 = 1.33Northwind typically leads the list when it’s mentioned at all.
Related: Visibility Score · Share of Voice
Citation metrics
Citation Rate
Unit: percentage (0-100%)
The percentage of AI answers that cite one of your pages as a source.
Why it matters - Mentions are about your brand name showing up. Citations are about your content being the source the AI pulled from. This is what GEO content strategy targets directly.
Formula
Citation Rate = answers_citing_you / total_answersInputs
| Input | Description |
|---|---|
answers_citing_you | Count of answers that include at least one citation pointing to a domain you own. |
total_answers | Count of all answers in the selected prompts and date range. |
Example
You’re Northwind (northwind.com), tracking Contoso and Fabrikam. Across two prompts asked on ChatGPT and Perplexity, you collect 4 answers with these citations:
| # | Prompt | Model | Citations |
|---|---|---|---|
| 1 | A | ChatGPT | northwind.com/features, contoso.com |
| 2 | A | Perplexity | contoso.com, northwind.com/features, fabrikam.com |
| 3 | B | ChatGPT | contoso.com/guide, fabrikam.com |
| 4 | B | Perplexity | northwind.com/tracking |
Northwind-owned pages appear in answers 1, 2, and 4 — 3 of the 4 answers.
Citation Rate = 3 / 4 = 75%Related: Citation Share · Visibility Score
Citation Share
Unit: percentage (0-100%)
Your citations as a percentage of all citations in the same AI answers.
Why it matters - If your Citation Rate is flat but Citation Share is dropping, competitors are winning more of the source slots.
Formula
Citation Share = your_citations / total_citationsInputs
| Input | Description |
|---|---|
your_citations | Count of citations to any domain you own. |
total_citations | Count of all citations in the same answer set, across every domain. |
Example
You’re Northwind (northwind.com), tracking Contoso and Fabrikam. Across two prompts asked on ChatGPT and Perplexity, you collect 4 answers with these citations:
| # | Prompt | Model | Citations |
|---|---|---|---|
| 1 | A | ChatGPT | northwind.com/features, contoso.com |
| 2 | A | Perplexity | contoso.com, northwind.com/features, fabrikam.com |
| 3 | B | ChatGPT | contoso.com/guide, fabrikam.com |
| 4 | B | Perplexity | northwind.com/tracking |
Count citations:
- Total citations: 2 + 3 + 2 + 1 = 8
- Northwind citations: 1 + 1 + 0 + 1 = 3
Citation Share = 3 / 8 = 37.5%Related: Citation Rate
AI Crawler metrics
Bot Requests
Unit: count
Total requests from AI bots (GPTBot, ClaudeBot, PerplexityBot, and others) to your site in the selected date range.
Why it matters - Crawls feed citations. No crawls means no citations. A drop in crawls is an early warning for citation drops weeks later.
The delta on the card compares the last 7 days against the 7 days before that, using your current bot and path filters.
Active Bots
Unit: count
The number of distinct AI bots that accessed your site in the selected range.
Why it matters - Shows how broadly AI platforms are picking you up. Heavy crawling from a single bot is a more fragile position than steady crawling from several.
Top Provider
Unit: provider name
The AI provider (OpenAI, Anthropic, Google, Perplexity, and others) that sent the most bot requests in range.
Why it matters - Shows where your crawl traffic is concentrated, which helps you prioritize content for the platforms that actually read it.
Top Page
Unit: URL path
The page path that received the most AI-bot requests.
Why it matters - Your highest-AI-attention page. Often the best candidate for structured data or a freshness update.
AI Referral metrics (GA4)
These metrics come from your connected Google Analytics 4 property and cover visits to your site from AI search engines.
AI Sessions
Unit: count
Visits to your site from AI search engines like ChatGPT, Gemini, and Perplexity, measured by GA4.
Why it matters - Turns visibility into actual traffic. If Visibility Score is climbing but AI Sessions aren’t, the mentions aren’t generating clicks.
Conversions
Unit: count
Conversion events from AI-referred visitors, using the conversion goals you’ve set up in GA4.
Why it matters - The bottom line. Only as accurate as the conversion goals configured in your GA4 property.
Conversion Rate
Unit: percentage
The percentage of AI-referred sessions that convert.
Formula
Conversion Rate = conversions / sessionsThe delta shown on the card is in percentage points (pp).
Engagement Rate
Unit: percentage
The percentage of AI-referred sessions that are “engaged” by GA4’s definition: the visitor viewed 2+ pages, converted, or spent 10+ seconds actively on the site.
Why it matters - Filters out low-intent traffic. High sessions with low engagement means AI is sending visitors who aren’t a fit.
Formula
Engagement Rate = engaged_sessions / sessionsBounce Rate
Unit: percentage
The percentage of sessions that aren’t engaged: no second page view, no conversion, and under 10 seconds of active time (GA4’s definition).
Formula
Bounce Rate = 1 - Engagement RateHigh Bounce Rate with long Avg. Duration often means visitors read the page but didn’t click further. Not necessarily a failure.
Avg. Duration
Unit: seconds
The average time between the first and last event in a session.