That is a good question.
In the sport of distance running, the "epo era" is like a great myth, which decades later still needs to be propped up by adding more unsupported allegations like European coaches/managers bringing their medicine to Africa, or the lack of blood testing (which only started *after* the EPO era) for a drug that was untestable everywhere in the 90's, and that marathon runners today have gotten better at hiding, or that the federations are allowing selected individuals and groups get away with it. Oh what a tangled web we've weaved.
While we do find fast times in the 90's and 00's, these times have not been clearly linked to EPO. The athletes linked to EPO have not run fast times "close to WR". To all those that have asserted "EPO leads to untouchable world records", they have largely failed to show any strong correlation between EPO and the fast times. The examples often presented as demonstrations of EPO effectiveness are both low in quality, and low in quantity. Sumgong's PB ranks her 27th of all time. Looking at the top-50 women marathon performers, we only have 3 that were "busted" for EPO. 3 out of 50 is an exception, not the general rule.
Compare this to cycling, where we have plenty of statistics like 7 out of the top 10 winners of every year were eventually linked to EPO use or blood doping, and the "clean" athlete remains the rare exception finishing 7th or 8th. In distance running, the incidence of EPO use, or blood doping, at the top seems to be a rare exception that we have to dig deep to find.
What is also strange, for a universal drug, touted to bring massive benefits, even for elites, is how selective great world leading performances were, largely favoring narrow populations in East Africa, the occasional North African, and a couple Spanish athletes able to set indoor European records, and one British woman. In an era where East Africans brought the 10K world record well below 26:30, the first non-African to break 27:00 was Solinsky, after the epo-era. This makes me wonder why all non-African nations were unable to leverage the powerfully effective benefit of EPO, something untestable in the 90's, and that Africans were allegedly able to hide, as late as Bekele in 2007, to be more competitive at any event from 800m to the marathon for two decades. They barely moved their own national records. Looking at all American, European, Australian, and all non-African performances, over all events for two decades, there were no massive progressions during "epo-era" for these nations. Looking at African performances, the case still needs to be made, that the performances were achievable only through EPO.
Trying to make the "epo era" hypothesis fit the observable results presents many contradictions, which require further and further creative "web weaving".