AI citation patterns shift sharply by search intent, Wix data shows

Wix Studio’s AI Search Lab did not stop at ranking content formats against each other. The same 75,000 AI answers and 1,056,727 citations it pulled from ChatGPT, Google AI Mode and Perplexity also carry a second, sharper cut: how citation share moves once a query’s intent, informational, commercial or transactional, is held constant. This site’s earlier report on citation rates by content format covered the format-only numbers, listicles at 21.9%, articles at 16.7%, product pages at 13.7%, as overall averages across every intent mixed together. The intent breakdown is the piece that number was hiding.

Articles lead on informational#

Averaged across all query types, Wix found articles cited in 16.7% of citations, second behind listicles at 21.9%. Split out to just informational queries, that share jumps to 45.48%, which the research states directly as articles being “cited 2.7x more for informational queries compared to other intents.” No other format in the dataset moves that far above its own baseline for a single intent category. Listicles, by contrast, sit close to their own average on informational queries, at 21.68%, meaning the informational advantage belongs specifically to articles, not to long-form content generally.

Lists and products win later#

Commercial queries, the research, comparison and best-of searches that precede a purchase decision, flip the leaderboard toward listicles. Their citation share reaches 40.86% there, which the study describes as “nearly double other intents,” up from a 21.9% baseline across all query types. Transactional queries, where the asker is closer to buying than comparing, hand the lead to product pages instead: a 24.88% citation share, up from a 13.7% baseline, the highest share product pages reach for any single intent. Read together, the three numbers describe a funnel: articles own the question stage, listicles own the comparison stage, product pages own the point of purchase.

Why we care#

A format-only reading of AI citation data, the kind this site already published, tells a publisher which format gets cited most often on average. It does not tell them which format to use for a specific piece of content they are about to write. This intent breakdown does: an informational how-to page structured as a listicle is not competing on listicles’ overall strength, it is competing against articles’ 2.7x advantage on informational queries specifically, on their own turf. The practical move is to match format to the intent behind the query a page is meant to answer, not to default to whichever format wins the leaderboard on average, since that average is really three different leaderboards averaged together.