Do ChatGPT and Google rank brands in the same order?

A comparison counts only when both engines directly ranked the same two brands in answers to the same prompt in the same run. ChatGPT Search and Google AI Mode put the same two brands in opposite order in 54,462 of 185,036 matched comparisons, or 29.4%. They kept the same order in 130,574 comparisons, or 70.6%.

One engine's ordered shortlist cannot stand in for the other engine's order. Compare the same prompt on both engines before treating a position as a shared market rank. BrightEdge reported query-level brand recommendation disagreement across three AI platforms. This study narrows the measurement to the relative order of brand pairs already shared by two engines.

At least one shared brand pair reversed in 27,309 of 50,620 comparable answer pairs, or 53.9%. A pair contained a median of three brand-pair comparisons. The remaining 23,311 answer pairs kept the same relative order for every shared pair.

The pair-by-pair rate adds up when an answer contains several shared brands. Audit the whole shortlist rather than relying on one cross-engine average. SparkToro and Gumshoe found that repeated runs rarely reproduced an entire ordered list. This result measures something different: two engines inside the same matched run.

When both engines chose the same top brand, 11,438 of 83,764 shared brand-pair comparisons reversed, or 13.7%. At least one reversal appeared in 6,469 of 21,476 same-top-brand answer pairs, or 30.1%. When the top brands differed, 38,683 of 89,083 comparisons reversed, or 43.4%.

First-place agreement is a useful result, but it does not prove that the rest of the shared shortlist has the same order. Measure the winner and the comparisons further down the list separately.

When both brands appeared in both engines' top three, 9,579 of 33,982 comparisons reversed, or 28.2%. The rate was 17,808 of 60,416, or 29.5%, inside both top fives and 27,075 of 90,638, or 29.9%, when at least one brand ranked sixth or lower.

The headline is not just a side effect of comparisons deep in the list. Teams should compare the order of leading brands as well as the first recommendation.

At least one reversal appeared in 7,258 of 22,332 answer pairs with two shared ranked brands, or 32.5%. The rate was 9,042 of 15,086, or 59.9%, with three; 9,207 of 11,296, or 81.5%, with four or five; and 1,802 of 1,906, or 94.5%, with six or more.

Any-reversal rates rise automatically with more possible comparisons. Compare answer pairs with similar shared-list sizes before treating one category or brand as unusually unstable.

Real Estate reversed 546 of 1,656 comparisons, or 33.0%, and Professional Services reversed 1,294 of 3,955, or 32.7%. Education had the lowest displayed rate at 563 of 2,190, or 25.7%. Every displayed industry contains at least 1,000 comparisons.

Industry changes the comparison baseline, but the observed rates do not show that industry caused an engine to reverse a brand pair. Use the industry breakdown to prioritize review, not to infer a ranking rule.

Ahrefs and Semrush reversed in 188 of 377 matched comparisons, or 49.9%. Asana and ClickUp had the most reversals among displayed pairs with 279 of 742, or 37.6%. Displayed names were merged to one standard form before publication.

Named pairs identify where to inspect the prompt and answer language. They do not show which engine's order was correct or whether either order affected buyer behavior.

The window produced 187,150 matched answer pairs. Exactly 50,620 contained at least two comparable shared ranked brands, while 136,530 did not. The final set contained no ties, no duplicate comparisons, no blank brand names, and no ranks below one.

The headline was 29.4% from June 1 through July 16, 31.3% with the first qualifying run per prompt, and 29.2% without merging brands into their parent brands. The study does not measure brands that one engine omitted, factual quality, recommendation accuracy, user demand, or causation. It measures relative order only where a direct comparison exists.

The engines reversed 29.4% of comparable shared brand pairs, and 53.9% of comparable answer pairs contained at least one reversal. Even same-top-brand answer pairs contained a reversal 30.1% of the time.

Track inclusion, first recommendation, and pair-by-pair order as separate measures. Run the same priority prompts on both engines. Review the underlying answer before averaging rank across engines or treating one screenshot as a market position. Rerun this method over the same window length next quarter before treating a change as real movement.