Ask AI for the best espresso machine and you get one view of the category. Ask about pods or a smaller budget and another comes into focus. For an espresso brand marketer, the difference matters: a broad category query can miss buying situations in which a product is a prominent option.
Gist studied home-espresso advice using configured ChatGPT and Perplexity APIs and Google AI Overviews. Breville led our balanced recommendation review, followed by De’Longhi and Gaggia. Nespresso’s presence was much more dependent on the question and engine.
The study covers 3,933 saved answers. We analyzed recorded names and source links across that archive, then reviewed the full saved text of a balanced selection of 177 answers to distinguish product recommendations from other mentions.
Breville leads on all three engines
Breville was offered as an explicit choice or conditional shortlist option in 92 of the 177 reviewed answers. De’Longhi followed with 65, then Gaggia with 35. Each engine put those three in the same order.
Answers recommending a brand
Explicit choices or conditional shortlists, out of 177 reviewed answers.
These are counts of reviewed answers recommending a brand, not shares of people or purchases. An answer can recommend several brands. The review includes purchases beyond the machine itself: Breville’s total comprises 91 answers offering a machine and one offering a grinder.
The leading three kept their positions when borderline recommendations were included and when any one question was removed. Their order holds under those checks; it is not a universal product ranking.
Nespresso barely registers on the broad question. Pods tell a different story.
Nespresso was named in none of the 20 saved ChatGPT answers to “What's the best espresso machine?” It appeared in two of Perplexity’s 24 answers to the same question.
On the explicit pod-machine question, ChatGPT named Nespresso in all eight saved answers. Perplexity named it in all twelve.
Nespresso in three buying situations
| Exact question | ChatGPT | Perplexity | Google AI Overviews |
|---|---|---|---|
| What's the best espresso machine? | 0 / 20 | 2 / 24 | 0 / 24 |
| Which pod espresso machines are worth buying for home use? | 8 / 8 | 12 / 12 | 16 / 17 |
| Which espresso machines under US$300 are worth considering? | 0 / 16 | 31 / 32 | 2 / 15 |
The budget question supplies the useful surprise. It never mentions pods, yet Perplexity named Nespresso in 31 of 32 answers. ChatGPT did not name it in any of its 16 answers.
In one checked Perplexity budget answer, the Vertuo Next was offered for convenience alongside machines from Breville, De’Longhi, Casabrews and Flair. The answer’s final advice conditionally favored the Breville Bambino, depending on the person’s grinder. Nespresso entered the decision without becoming its universal winner. This describes the saved advice; it is not an endorsement of its prices or specifications.
The engines also turn to different sources
The leading cited host was breville.com in ChatGPT, goodhousekeeping.com in Perplexity and youtube.com in AI Overviews. Their respective citation rates were 33.7%, 42.5% and 68.0%, after giving each question equal weight within its engine.
Leading cited host on each engine
YouTube remained the leading AI Overviews source after identical texts within each question were merged. But a recorded link does not establish which page shaped a recommendation. A manufacturer, a publisher and a video platform also play different roles; treating their citations as interchangeable endorsements would lose that context.
Read the shortlist with the question beside it
Breville’s lead and Nespresso’s uneven presence are both part of the category picture. A useful visibility report shows the overall pattern and the buying situations underneath it. It also keeps three signals separate: being named, being recommended and appearing as a cited source.
The full report follows those distinctions through the brand results, engine differences, source analysis and individual answers. It includes every measured question and the evidence limits behind the findings.
About the research
Gist collected these answers on 28 October 2026 (UTC) using ChatGPT chat-latest, Perplexity sonar and Google AI Overviews through its native provider route. Two 30-entry panels shared one exact question, producing 59 unique questions, 60 panel entries and 177 completed question–engine runs. Sampling was adaptive, so the 3,933 nonempty answers are not equal-sized or necessarily independent samples.
Three AI reviewers coded separate engine partitions of the balanced 177-answer selection; 12 answers received a targeted second-reader check. Four reviewed AI Overviews have unresolved product identities. Identifiable advice was retained, but missing names were not reconstructed. Failed attempts and successful searches without an overview are excluded from the answer denominator.
The second panel included editorial amendments, including pod coverage. The study does not establish a causal winner between question-development methods, consumer demand, sales effects or product-test results. The report contains the complete methods and qualifications.