How ChatGPT Recommends DuckDB

Across 100 data warehouse and ETL tools questions, DuckDB appeared in 13 answers (tied for rank 10 of every brand named), was recommended in 13 (tied for rank 10) and was ChatGPT's own pick in 9 (tied for rank 10). That is a pick rate of 69% of its appearances, below the 12-brand median of 90% for brands named 10 or more times in this teardown.

When ChatGPT picked DuckDB, it named it first among the brands it picked in 2 of those 9 answers. On the 20 direct advice questions ("which should I use"), ChatGPT never recommended or picked it.

These are the labels ChatGPT attached to DuckDB in "best for" table columns and scenario lists, quoted as written. We found 11 such labels for DuckDB in this category; the most frequent are below.

• "💻 Personal projects & analytics"
• "Local/embedded analytics"
• "Local analytics, notebooks, small teams"
• "Need cheap local analytics"
• "Small-to-medium datasets, local analytics"
• "Small-medium workloads, ETL, local analytics"

If this is not how you describe your product, that gap is the first thing to fix: AI answers repeat the positioning they find most often across the sources they read.

1 comparison question named DuckDB against at least one rival. ChatGPT picked both DuckDB and at least one named rival in 1, only DuckDB in 0, and only a rival in 0. In that question ChatGPT made both DuckDB and a rival its picks instead of declaring a single winner.

The 9 answers that picked DuckDB carried 25 source citations in total. The most-cited domains in those answers:

• clickhouse.com (vendor site): 6
• google.com (vendor site): 3
• github.com (other third party): 2
• snowflake.com (vendor site): 2
• amazon.com (vendor site): 2
• apache.org (vendor site): 1

Across all 13 answers that recommended DuckDB, picked or not, there were 38 citations.

DuckDB's own site (duckdb.org) was cited once across all 100 answers in the category, in 1 answer.

Google check: for 0 of the 9 questions where ChatGPT picked DuckDB, duckdb.org also ranked in Google's top ten organic results for the same question. So most of these picks came on questions where its own site did not rank.

The brands that appeared most often in the same answers as DuckDB, and in how many of those answers ChatGPT picked them. This is the competitive set ChatGPT has built for it, which is not always the one a vendor would choose.

• ClickHouse: in 12 of DuckDB's 13 answers, picked in 12
• Trino: in 7 of DuckDB's 13 answers, picked in 7
• BigQuery: in 6 of DuckDB's 13 answers, picked in 5
• Amazon Redshift: in 5 of DuckDB's 13 answers, picked in 4
• Databricks: in 5 of DuckDB's 13 answers, picked in 4
• Snowflake: in 5 of DuckDB's 13 answers, picked in 3

DuckDB is on ChatGPT's data warehouse and ETL tools shortlist but is not one of its top three picks: it was recommended in 13 answers and picked in 9, against 46 picks for Snowflake, the most-picked brand. Its lowest pick rate is on direct advice questions, where ChatGPT picked it in 0 of 20.

In 9 of the 9 answers that picked DuckDB, duckdb.org did not rank in Google's top ten for the same question, so Google rankings alone do not explain where ChatGPT sends buyers.

Finding which sources shape the answer in a category, and making sure they describe the brand accurately, is the core of AI SEO.

Source: Teardown 13: data warehouse and ETL tools. Each teardown runs 100 buyer questions (category, head-to-head, alternatives and advice phrasing) through ChatGPT with web search on, logged out, United States, English, one run per question.

Every answer was coded under codebook v2.0 by two independent AI readers working blind from the published reader protocol. Each reader coded every brand mention as picked, recommended, listed, passing or anchor; where the two disagreed, a third AI reader settled the code. Across all answers coded under v2.0, the two readers agreed on 97.8% of brand codes.

"Picked" means the answer made the brand its own verdict: "my pick", "choose X if", a shortlist it tells the buyer to act on, the #1 of its ranking, or the direct answer to a "which should I use" question. "Recommended" means the answer assigned the brand to a stated case or fit ("best for small teams") or picked it, so every picked brand also counts as recommended. "Appeared" counts answers where the brand was recommended or listed. Every count is a count of answers. Use-case labels are quoted from the answers as written. Google checks use the top ten organic results for the same question collected the same day.

Limits: one run per question, one country, one point in time. AI answers vary between runs, so treat these as frequencies for this sample, not fixed positions. High Salience has no commercial relationship with any brand on this page. If you work for this brand and see an error, write to info@highsalience.com and we will check it against the dataset.

Full per-answer datasets (CSV) are on each teardown page. All brand pages.

Does ChatGPT recommend DuckDB?

In our data warehouse and ETL tools teardown (web search on, United States, September 26, 2026), ChatGPT named DuckDB in 13 of 100 buyer questions, recommended it in 13 and made it its own pick in 9.

What does ChatGPT say DuckDB is best for?

The label ChatGPT used most often for DuckDB was "💻 Personal projects & analytics". Other labels included "Local/embedded analytics", "Local analytics, notebooks, small teams", and "Need cheap local analytics".

Which brands does ChatGPT compare DuckDB with?

In data warehouse and ETL tools answers, DuckDB appeared most often alongside ClickHouse, Trino, BigQuery, Amazon Redshift, Databricks, and Snowflake.

The Category Salience Brief runs your category's buyer questions across Google and the AI surfaces, shows how each answer positions you against competitors, and ranks what to fix first.