Unless you are really limited before, speedtest won't show a difference. Most people's internet connections are a lot slower than their WLAN speeds.
In my experience, the difference between the lower 5GHz channels and the upper ones on routers that are pre-FCC regulatory changes is about 3-4dB, IE only about a 2x difference in signal strength. Most routers do NOT broadcast at the full FCC limits and all clients are limited to between 25-50mw anyway, so you'd have issues if your router was 20x stronger (IE the client can hear the router, but it cannot be heard in return, so the connection would "be there" but it wouldn't work).
In general most routers broadcast around 80-120mw, which means that the lower power on the low 5GHz channels really is only about half the broadcast power of the upper ones.
Of course doubling up the broadcast power is still something nice.
In testing my Archer C8 at a fixed position roughly 12ft from the router, line of sight choosing channel 148 my average transfer speed from the router to my laptop is 58MB/sec. If I choose channel 36, the average transfer speed drops to 54MB/sec. Granted, that is less than a 10% difference, but at extreme range it is a little more apparent. 2/3rds of the way across my house and a floor up, I can get 10MB/sec down choosing channel 148, but on channel 36 it is only 8MB/sec.
Get the laptop about 5ft from the router and line of sight and there is effectively no difference between channels (all selections result in roughly 61MB/sec).
So in general, if you can and there is no interference, select channel 148+. If there is, go the lower channels. There is some difference, but it isn't night and day.
Another example if it helps, my outdoor router (only the antennas are outdoors), a WDR3600 in testing on 5GHz at a distance of 30ft line-of-sight I can get around 15MB/sec with the upper channels and around 13MB/sec on the lower channels. At 70ft line-of-sight it is about 11MB/sec upper channels and 8MB/sec lower channels. At 15ft line-of-sight both channel selections result in 25MB/sec.
Of course if it is "free", take the extra performance. In my house I have two routers running on opposite sides, but if do co-channel on both there IS interference, though small. Setting up my laptop and my tablet hammering a file transfer on both placed about 10 feet apart in the middle of my house (equidistant from the routers, though obviously each client is closer to the router it is connected to) I see about a 25% performance hit on both instead of choosing the lower 5GHz on one router and the upper 5GHz on the other router (zero impact then). Now if the clients are on opposite sides of the house, there is effectively no interference when doing the same test.
Now if you have a post-FCC regulatory change, there isn't really a concern as the power levels are the same on both parts of the band (now if DFS channel selection was common. Sigh).