Do AI assistants like ChatGPT and Perplexity choose to quote “contemporary” content material? We analyzed 17 million citations throughout 7 AI search platforms to seek out out.
As we clamor to optimize for AI search visibility, I’ve seen recommendations that we should always replace pages extra steadily to spice up their likelihood of being cited.
To check the speculation, we used knowledge from Ahrefs Model Radar to extract 16.975 million cited URLs from throughout ChatGPT, Perplexity, Gemini, Copilot, AI Overviews, and the great old school natural Google SERPs. We then calculated…
- The typical variety of days for the reason that cited articles have been printed (utilizing the date the articles have been first seen by our crawler), and
- The typical variety of days for the reason that cited articles have been final up to date (utilizing the final printed date from our Content material Explorer database)
…to see if AI assistants actually “choose” brisker content material when in comparison with conventional Google Search.
See your AI visibility with Ahrefs Model Radar
The info for this analysis comes from Ahrefs’ Model Radar. Uncover what AI assistants say about your model, attain new audiences, and switch AI search into your subsequent progress channel.
- In comparison with conventional search outcomes, AI assistants choose citing brisker content material.
- The typical age of URLs cited by AI assistants is 1064 days, in comparison with 1432 days for URLs in natural SERPs—25.7% “brisker”.
- Google’s AI Overviews and natural search outcomes are the almost certainly to quote older pages.
- ChatGPT is almost certainly to quote newer pages.
- Perplexity and ChatGPT order their in-text references from latest to oldest.
If we common the age of cited URLs in all positions throughout ChatGPT, Perplexity, Copilot and Gemini, we get a mean time since publication of 1064 days, or 2.9 years. For the natural SERP, the typical time since publication is 1432 days, or 3.9 years—one full yr older.
this alone, it appears that evidently AI assistants have a transparent desire for newer URLs, with cited URLs being, on common, 25.7% “brisker” than URLs in natural search outcomes ((1432-1064)/1432).
Wanting on the common time elapsed since articles have been final up to date, we discover a comparable story.
AI assistants present a desire for extra not too long ago up to date content material, with a mean time elapsed since final replace of 909 days, in contrast with 1047 days for the natural SERP—a smaller distinction of 13.1% ((1046-909)/1046).
If we break the info out, we get a clearer thought of the “freshness” desire of every AI assistant and search platform.
Google’s prime three AI Overview citations present the strongest desire for older content material, with the typical cited article being over 1432 days (3.9 years) previous, and final up to date 1047 days (2.9 years) in the past—very near the typical age of URLs in Google’s natural SERP.
Perplexity’s in-text citations present the following strongest desire for older content material (1166 days), adopted by Gemini (1118 days), then Copilot (1056 days), then ChatGPT’s in-text references (1023 days) and citations (958 days).
Right here’s the info in desk kind:
| AI search platform | Common days since publication | Common days since final up to date |
|---|---|---|
| Google AIO (prime 3) | 1432 | 1067 |
| Natural SERP | 1416 | 1047 |
| Perplexity (textual content order) | 1166 | 993 |
| Gemini | 1118 | 831 |
| Copilot | 1056 | 865 |
| ChatGPT (references) | 1023 | 865 |
| ChatGPT (citations) | 958 | 989 |
If we use the natural search outcomes as a baseline, we will extra simply see the extent to which every AI assistant prefers to quote newer content material.
Specializing in the times elapsed since publishing, ChatGPT reveals the strongest desire for brand spanking new content material, preferring to quote URLs which are 393 days newer than natural Google leads to their in-text references, and a whopping 458 days newer of their citations.
In comparison with the natural SERP, all AI assistants present a transparent desire for brisker content material—besides Google’s AI Overviews, which cites content material that’s 16 days older, on common.
Right here’s the knowledge:
| AI search platform | Distinction from natural SERP (printed) | Distinction from natural SERP (up to date) |
|---|---|---|
| Google AIO (prime 3) | 16 | 21 |
| Natural SERP | 0 | 0 |
| Perplexity (textual content order) | -250 | -53 |
| Gemini | -298 | -216 |
| Copilot | -360 | -182 |
| ChatGPT (references) | -393 | -182 |
| ChatGPT (citations) | -458 | -58 |
We will additionally see how the age of cited URLs varies by their place. Right here’s the graph exhibiting the typical time since publication for positions 1 by means of 10.
The natural SERP outcomes recommend that older outcomes usually tend to seem in high-ranking positions, whereas the AI assistants every present a comparatively flat pattern, with most cited URLs being comparatively shut in age.
Wanting on the common time since cited articles have been final up to date, each ChatGPT’s and Perplexity’s in-text references comply with the identical quotation sample, seemingly ordering references from latest to oldest.
This sample is clearer if we have a look at the correlation between the order of cited URLs and their age (excluding AI Overviews).
Most platforms present a adverse correlation between quotation and content material age, suggesting that older content material is often cited first (and ranked larger within the natural SERP).
However Perplexity and ChatGPT’s references each present a optimistic correlation between quotation order and content material age, with newer content material being cited first. This looks as if intentional age-related ordering of their in-text references.
Ultimate ideas
When in comparison with conventional search outcomes, AI assistants appear to have a desire for citing newer, brisker content material. To enhance your LLM visibility, there could also be some profit to commonly updating content material.
However I can’t say this with out issuing just a few caveats and cautions:
- Google is the least influenced by content material freshness, and Google remains to be the place nearly all of searches occur.
- Google’s John Mueller has warned in opposition to updating publish dates with none corresponding modifications within the web page content material.
- Content material freshness is one amongst many components that affect visibility. Low-quality, irrelevant content material that’s up to date every single day is not going to have a magic optimistic impact in your visibility in AI search.
- Each selection has a chance value, and I believe that almost all manufacturers will see higher outcomes from creating new, high-quality content material than they are going to from committing sources to extraordinarily frequent content material updating.
Feedback or suggestions on this analysis? Let me know on LinkedIn.
