On the sales numbers figures... 3 million for Civ4 is through 2008.
SteamDB, citing SteamSpy, estimates 1.3 million copies on Steam, most of which would have been post-2008, with another site estimating 3 million on Steam, and there is also GOG, and some amount of post-2008 disc sales. So versus V's 8 million figure, I think it's safe to say IV has sold more than half as many. And, as Fippy notes, that was against a considerably smaller potential audience. Has the increased unit sales of V and VI really reflected higher popularity of the game, or just the expansion of the video game market in general? I'm not sure, but I'm skeptical that "1 UPT increased sales" is a correct takeaway.
I like Marla_Singer's analysis of hexes versus squares as well. I've long thought that the 1.4x benefit cited by hex advocates is overstated in strategy games like Civ, but the restriction in movement choices due to the fewer adjacent tiles is interesting. I would agree that the fewer-adjacent-tiles exacerbates the 1 UPT problem. Get rid of 1 UPT, and I'm not sure I would have a significant preference either way for hexes versus squares versus isometric. I suppose visually, I find the imperfectly-geometric square/isometric graphics in Civ4/3 (notably islands/coasts) to be preferable to the overly-geometric hexagonal edges of V/VI tiles/islands, as it just doesn't look naturally seeing, say, Crete as three perfect hexagons next to each other, but that doesn't affect gameplay.
And while I'd prefer not to, I have to agree with Marla_Singer that given that Ed Beach is still the lead designer, a significant change in design for Civ VII seems unlikely. Maybe we'll be pleasantly surprised, but considering that VI was largely consistent from V, and sold well, why would they rock the boat? Although one could argue that they didn't have a great need to rock the boat from IV to V, either.
I'll admit that I liked VI when I finally tried it - the districts idea felt fresh, and it has more of a sense of empire than V, particularly in Vanilla where you can rapidly expand without worrying about cities flipping. But the more I've played it, the more I've suspected that what made it fun initially was the novelty factor. The cogs of the Civ machine all move so slowly compared to III and IV. District improvements that give a handful of production/commerce/science often take many turns to build. Units are produced slowly, because otherwise the 1 UPT limit would be more of a problem than it is. AI turn times seem to take a while even on Standard size maps, but the AI was better in III and IV and quicker in IV (and sometimes in III depending on the settings). Government options do move quickly, but there are so many choices that it slows back down when deciding whether to tweak the government every few turns, rather than simply deciding on a government or a handful of civics like in III/IV.
1 UPT combined with the scale - fairly tightly packed cities - is, IMO, the fatal flaw in later games in the series.
Old World works despite 1 UPT because its scale is vastly different. The subdividing tiles idea is interesting. But ultimately, I think Fippy is right too - IV already had incentives to not infinitely stack, and perhaps hit the sweet spot in that regard.