Nostos GEO
FAQ · Small websites

How can a small website appear inside ChatGPT or Perplexity answers?

By Fabián Torres · November 5, 2026 · 2 min read

First, allow access to each AI's search crawler. Then publish pages that answer one specific question and include sources, quotations and verifiable data. Being small doesn't rule a site out: in one experiment, the lowest-ranked pages gained the most from these changes. None of this guarantees you will appear.

115.1%
visibility increase for pages ranked fifth in search results after adding citations to sources. The top-ranked page lost 30.3% on average.
Source: Aggarwal et al., KDD 2024. Experiment with the authors' own generative engine, on pages already in Google's top five.
"Notably, lower-ranked websites, which typically struggle for visibility, benefit significantly more from GEO."— Aggarwal et al., KDD 2024.

What does a small website need to do?

Four things, in this order. To see the whole path, check the stage map.

1. Can crawlers access your site?

They need to. OpenAI says sites that block OAI-SearchBot are not shown in ChatGPT search answers, though they can still appear as navigational links. Perplexity recommends allowing PerplexityBot so its results can surface and link your site. Check your robots.txt file and your firewall too.

2. Does each page answer one specific question?

It should. Put the answer first, then explain it. A page that solves a precise problem is easier to use than one that covers the topic in general.

This especially suits a niche site, as we explain in Why shouldn't a hardware store compete with Wikipedia?

3. Does your text have sources, quotations and data?

It helps if it does. In the same experiment, citing sources, adding quotations and including statistics improved visibility by 30% to 40%. Tested on Perplexity, adding quotations gave a 22% improvement. Use data you can back with a source.

To understand what AI does with those pages, read how they decide which sites to cite.

4. Does repeating keywords help?

No. Keyword stuffing barely helped in the experiment, and on Perplexity it performed 10% worse than no optimization.

How far is a typical site?

Quite far, according to our measurement. In July 2026 we used Nostos to measure 39 Argentine e-commerce domains. The average was 35 out of 100 and the highest score was 62. 23% had a technical markup score of zero, and 62% scored below 30 on content structure. The sample leans toward large operators, and these areas were still weak.

How can you find out where your site stands?

To find out where your site stands with AIs, analyze it with Nostos: the result shows you what to fix first.

What are the limits of this answer?

Several. The experiment dates from 2024 and tested pages that had already been retrieved; it doesn't measure how to get retrieved, and it didn't use today's ChatGPT. The authors also didn't evaluate the effect on search rankings.

Analyze your site for free

Get your GEO score across the eight criteria backed by academic research.

Run the analysis at nostosgeo.app