Three sites generated 215,128 “best software” pages — and Perplexity cites them

On 2 September 2026 the research group Trellner published report TR-2026-009, which traces which documents sit behind AI-generated software recommendations. Of 7,534 citations returned by two Perplexity models, 59.8% point to domains ranked worse than #100,000 in the Tranco top-1M list, and 23.4% are not in the top million at all.

How the study was run

The researchers put 380 buyer-intent categories — from CRM software to museum collection management software — to perplexity/sonar and perplexity/sonar-pro through OpenRouter, one prompt per category per model, 760 calls in all. Each call asked for a ranked top five as JSON with each product's official homepage domain. All 760 returned a parseable answer, producing 3,800 recommendation slots naming 1,807 distinct products and 7,534 citations spanning 2,055 distinct domains. Google was left out of the study.

What the citation mix shows

The median Tranco rank of the 5,768 citations pointing at ranked domains is 71,611. Concentration at the top is unremarkable: the ten most-cited domains take 17.3% of citations. g2.com leads with 291, followed by reddit.com with 261 and guideflow.com with 194 — a vendor of interactive product demos that competes in none of the categories asked about, yet whose blog was cited across 96 of the 380 categories. Gartner follows on 158. Wikipedia was cited three times in 7,534.

Three sites, 215,128 pages

wifitalents.com, worldmetrics.org and gitnux.org look like one operation. All three were registered through NameCheap between December 2023 and May 2024, delegate DNS to the same Cloudflare nameserver pair, and run the same page template. Their sitemaps list 103,578, 107,083 and 105,541 URLs, of which 70,731, 71,684 and 72,713 are /best/…-software/ pages: 215,128 generated buying guides across three brands. None of the three domains existed before December 2023.

Two of them give their homepage the HTML title “Facts & Grounding Page” — grounding being the retrieval step in which a model fetches documents to condition an answer on. For the same category, “project estimation software”, each brand ranks a different winner and credits three named staff: nine distinct people for one question.

Limits of the finding

The result covers Perplexity only. The two model tiers share a retrieval layer — 289 of 380 categories returned byte-identical citation lists — so they should be read as one search stack sampled twice. Tranco rank is a popularity measure, not a quality one, and the authors stress that a low rank is not in itself an accusation.