How many tracked queries saw the leader change at least once over the study window

The week-to-week leader change rate (56.4%, covered on its own page) tells you how often the top company changes between any two adjacent weeks, averaged across the whole study. A different, complementary question is: over the full length of the study, how many of the tracked query-engine combinations ever experienced a change at all, even once.

This study tracked 480 series, one for each combination of prompt and engine, each with a median of 13 weekly observations across the run window from 2026-04-27 to 2026-08-03. Of those 480 series, 395 saw the identity of the top-mentioned company change at least once somewhere across the window, 82.3%. The remaining series kept the same leader throughout, a much smaller share, 85 series out of 480, held one company in first place for every single observed week.

Why this number is more useful than it first appears

The week-to-week rate can be misleadingly reassuring if read the wrong way. A 56.4% chance of change in any given week sounds like the leader is stable more often than not, in any one comparison. But stack enough weekly comparisons together, over a median of 13 weeks, and the odds that a series goes the whole distance without a single change fall quickly. That is exactly what the data shows: 82.3% experienced a change eventually, even though any single week-to-week comparison was closer to a coin flip.

This is the same logic as a coin that comes up different roughly half the time on any single flip: over many flips in a row, the chance of never once flipping to the other side becomes small. A company should expect, over any multi-month tracking period, that its position will be displaced at least once even if its underlying market strength is not declining.

What distinguishes the small stable minority

Only 85 series out of 480 held a single unchanged leader the entire window. This dataset does not break out what those series have in common, whether they cluster in particular categories, particular prompt types, or particular companies, and that is a genuine limit of what can be said from the published aggregates. A plausible hypothesis is that these represent queries with an unusually dominant, well-established single answer (a category with one overwhelmingly prominent company), but confirming that would require inspecting the specific companies and prompts behind those 85 series directly, which is possible from the underlying data but was not part of the aggregate figures reported here.

What a reader should do with this number

If a company's own tracked series has gone unchanged for several weeks, this figure (82.3% eventually see change) is a caution against assuming that streak will continue indefinitely. Conversely, a single change after a long stable run is consistent with ordinary background turnover and is not automatically evidence of a real shift in market standing.