Jacksonville, FL is an extreme example of this. Jacksonville's municipal boundaries are huge - it covers 875 sq. miles[1]. This is ~1.85x the area of New York City[2], and 3.77x the area of San Francisco[3].
Within the US, it's preferable to use core-based statistical areas (CBSAs)[4] for this type of analysis. They're defined by commuting zones, and they're not sensitive to political boundaries except to the extent that political boundaries actually affect commuting patterns. If you want to limit yourself to the "dense" part of the metro area then you can use the Census-defined Urbanized Areas[5], but the definitions of those might not match your intuitions about what should be included.
Assuming we use CBSA population, then the box that contains Jacksonville should instead have Atlanta, which has an CBSA population of ~5.8 million vs. only ~1.5 million in Jacksonville[5].
The statistical agencies for other countries have similar concepts. For Europe, Eurostat defines metropolitan regions[6], which have a similar definition. The OECD also maintains a data set of population by metropolitan area for OECD countries, although the methodology isn't consistent across all countries in that data set[7].
TL;DR - political and municipal boundaries aren't comparable across regions, use commuting zones or density-based region definitions instead. If you see Jacksonville, FL in a list of "biggest cities", it's always a red flag that they're using municipal boundaries.
[1] https://en.wikipedia.org/wiki/Jacksonville,_Florida
[2] https://en.wikipedia.org/wiki/New_york_city
[3] https://en.wikipedia.org/wiki/San_Francisco
[4] https://en.wikipedia.org/wiki/Core-based_statistical_area
[5] https://www.census.gov/programs-surveys/geography/guidance/g...
[5] https://data.census.gov/cedsci/table?q=United%20States&tid=A...
[6] https://ec.europa.eu/eurostat/web/metropolitan-regions/data/...