Why it matters
Not all mentions are equal. When an AI lists ten brands in its answer, the first three get disproportionate attention from the reader — both in chat interfaces where users rarely scroll and in citation panels where top results get most clicks. A brand with a mention rate of 40% and an average position of 2.1 has materially better visibility than a brand with a 40% mention rate and an average position of 8.5. Average position also correlates with perceived authority. Readers interpret the first-mentioned brand as the AI’s primary recommendation, even when the AI does not explicitly rank its suggestions. Moving from position 6 to position 3 often produces larger downstream effects than moving from a 25% to a 30% mention rate.Formula
position_in_response— the rank (1 = first) at which the brand appears in each AI responsecount(responses_with_mention)— total number of responses where the brand was mentioned
Example calculation
A travel bag brand is mentioned in 18 responses across 50 tracked prompts. Its positions in those 18 responses are: 1, 2, 2, 3, 3, 3, 4, 4, 5, 5, 5, 6, 6, 7, 7, 8, 9, 10.Cited Index benchmark — average position
The following percentiles are computed from the Cited Index, March 2026 edition. Lower is better: a p25 of 4.1 means that 25% of brands have an average position of 4.1 or better when mentioned. These benchmarks reflect non-branded category prompts only. Your own dashboard may show different position distributions because your prompt library includes a broader mix of prompt types.Overall distribution
Across 281 brand-category observations:| Percentile | Average position |
|---|---|
| p10 | 3.3 |
| p25 | 4.1 |
| p50 (median) | 5.6 |
| p75 | 7.3 |
| p90 | 10.0 |
By category
Consumer categories cluster near the top of the list with tight position ranges, while B2B and fintech categories show wider spreads — Audio & Wearables brands average position 4.3, while CRM & Sales brands average 6.6.| Category | Brands | p25 | p50 | p75 |
|---|---|---|---|---|
| Audio & Wearables | 22 | 3.5 | 4.3 | 5.5 |
| Skincare & Beauty | 21 | 3.7 | 4.4 | 5.1 |
| Travel & Luggage | 21 | 4.0 | 4.5 | 5.2 |
| Digital Payments | 55 | 3.8 | 5.6 | 7.7 |
| Online Learning | 34 | 4.2 | 5.6 | 7.5 |
| HR & Payroll | 31 | 4.1 | 6.0 | 6.9 |
| CRM & Sales | 55 | 5.0 | 6.6 | 8.9 |
| Credit Cards | 42 | 5.5 | 6.6 | 8.8 |
What changes this metric
- Editorial coverage in authoritative sources — when LLMs cite your brand from a high-authority source, you tend to appear earlier in the response
- Content specificity — brand content that directly answers the prompt (comparison pages, feature breakdowns) gets surfaced earlier than generic marketing pages
- Recency — recent content and recent mentions in fresh training data tend to appear in earlier positions
- Structured data — brands with clean schema markup tend to be parsed into earlier positions in AI responses, especially on Perplexity
- Prompt intent alignment — if the prompt is highly specific to your product, you appear earlier; generic prompts surface brands in a less differentiated order
How Cited measures it
Cited parses each AI response for all brand mentions in order of appearance — first brand mentioned is position 1, second is position 2, and so on. See How we extract mentions for the full extraction layer.Related concepts
- Mention rate — whether your brand appears in the response at all
- Share of voice — your share of total mentions in the category
- How Perplexity ranks sources — the primary platform for position measurement
Frequently asked questions
Why is lower better for average position?
Why is lower better for average position?
Position 1 is the first brand mentioned or cited in an AI response. Position 10 is the tenth. Readers pay more attention to earlier items, so brands appearing in lower-numbered positions get more downstream engagement. Average position of 3.0 is better than average position of 7.0.
How is average position different from mention rate?
How is average position different from mention rate?
Mention rate measures how often you appear at all. Average position measures where you appear when you do appear. A brand mentioned rarely but always first (mention rate 20%, avg position 1.5) is not directly comparable to a brand mentioned often but always last (mention rate 60%, avg position 8.0). They measure different dimensions of visibility. The Cited Index benchmarks for both metrics are based on non-branded category prompts.
Does average position apply to all AI platforms?
Does average position apply to all AI platforms?
It is most reliable on Perplexity, which provides explicit source citations in ranked order. ChatGPT, Gemini, and Claude mention brands in prose, so position is inferred from text order rather than an explicit ranking. Cited tracks position across all platforms but position signal is strongest on Perplexity.
Why is the p10 for average position 3.3 and not 1.0?
Why is the p10 for average position 3.3 and not 1.0?
Averages smooth out individual rankings. A brand that appears in positions 1, 2, 3, 4, and 6 has an average position of 3.2 — still top-decile. A brand that always appears in position 1 (average 1.0) is rare, because most brands have variance across prompts and platforms. The p10 of 3.3 means the top 10% of brands consistently appear in the top 3–4 positions on average.
How do I improve my average position?
How do I improve my average position?
Earn editorial coverage in sources the AI considers authoritative (this moves you earlier in citation-based responses), create content that directly answers the prompts you want to rank for, ensure your schema markup is clean, and maintain freshness — LLMs favor recently-updated sources. See the Improve your Perplexity citations playbook for detailed tactics.