Many / most sources credit Westinghouse's experimental station, KDKA in Pittsburgh, as the first commercial radio station,
That's just Westinghouse propaganda;
AMRAD's experimental station, 1XE, was already broadcasting in 1917, and there were some others as well - there are some in California that have a better claim. If AMRAD continued to function today or if another company, like GE, purchased AMRAD early on, than 1XE and its successors would be seen as the first commercial radio station.
Having said that:
I recall reading years ago about experimental voice and music broadcasts as early as 1915, and it's well established that wireless Morse code communications were used in 1914 during the war. Thus: could the interval between Marconi's transatlantic experiment in 1901 and KDKA first broadcasting be shortened reasonably, and if so, by how much?
There was even Reginald Fessenden's famous 1906 experimental broadcast of
O Holy Night and other holiday music from a transmitter in MA, so pushing radio broadcasting sooner rather than later is possible. One thing to keep in mind is that non-commercial and educational stations outnumbered the commercial ones in the early years and continued all the way to General Order 40, which deliberately sought to eliminate all forms of non-commercial media from the airwaves (which didn't always work out as planned, as is the case with
WHA, the flagship of Wisconsin's public radio network). Even before General Order 40, many Americans who had an opinion on the subject would prefer anything other than commercial means to fund broadcasting. So one main thing should be that expanding commercial broadcasting earlier should not come at the expense of non-commercial media. The trick for expanding commercial broadcasting, and doing it earlier, is basically keeping them all at the "experimental" stage, in which case many more stations would be recognized as the first (
KCBS, for example, which started broadcasting just a mere couple of years after Fessenden's experiment) and maybe even other stations ITTL could take Fessenden's idea and run with it.