I think there's a factor that was touched on earlier in the discussion but not elaborated on: this was an ambitious, risky project from the beginning. The project to create a new intellectual property that would form the basis of multiple games is a big one. It's one reason why many Hollywood studios, which are typically headed by risk-averse business types rather than dream-big creative types, prefer to mine popular existing worlds for stories than to create new ones (another being, as pointed out earlier, that audiences tend to want the right mix of new content and familiar stories, worlds, or characters). I would argue that the Eora Project (a name I invented just now for lack of anything more real) was considerably more ambitious.
Consider: not only did Obsidian leave behind a lot of the familiar fantasy tropes that one gets from a standard, medieval Europe type of fantasy setting. They also made a lot of narrative work for themselves: animancy, firearms, deep backstory based on a history going back thousdands of years, and the rather novel notion that the gods of this world had been very literally created by people who had been disappointed to find that there was no architecture behind their world*. This work isn't just for them, it's for their players: to read, to absorb, to embrace. I was very skeptical of firearms and animancy coming into PoE, and Obsidian did a spectacular job of layering magic and divinity alongside chemistry and spiritualism in a seamless, nuanced way that put my concerns to rest quickly and quietly. On top of all of this, Obsidian made a very clear decision to eschew tropes, clichés, and standard fantasy props in favor of subtlety, nuance, and complex human interactions. No orcs. No good/evil axis. No Mary Sue character written in to tell the player how to think about their decisions. Even the history and the geopolitical environment attempted to have the texture and depth of a real world. Instead of "a nation of vikings who worship the god of war" or whatever, you have trading consortia and muddled interests and the very plausible consequences of having real, embodied gods who can act to change the world.
On top of the narrative and world-building ambitions, they decided to try to fix the obvious problems with the Infinity Engine games' mashup of a real-time with pause combat system bolted to a very turn-based D&D game system, not to mention the entrenched balance problems of D&D. So they set out to build their own mechanics where each stat has roughly equal value, each class is equally viable, and there are no trap options. Although they fell short of total success here (and who wouldn't?), they came close enough to their objectives that they reached what I would call a very good result. This, however, came with the cost to the player of a new game system to learn, one with nuances specific to RTwP gameplay, and a more complex build environment to manage, which is overhead when considering a new RPG. As others have mentioned, time economy changes as we age. I bounced hard off PoE and didn't return until White March 2 was released - as a kid with far fewer games to choose from I wouldn't have, but as an adult with limited gaming time and way too many game options, I can't always come home at the end of a day of systems analysis and sign up for some more, voluntary, unpaid systems analysis. When I came back, many changes had been made and there was more guidance on the boards and I dove deep into PoE and am very glad I did.
All of this to say that I think this was a massively ambitious project from the outset. Obsidian took on a lot of voluntary challenges in building the Eora world, and staked out some very bold narrative and mechanical ground. I would argue that PoE came out at the right time, and, as others have guessed, I guess that it may have oversold on a bit of a wave of nostalgia. But I am speculating here based on very thin evidence and I haven't looked at the release environment to validate my suspicions about nostalgia - I am mostly working off of my own memory of salivating while waiting for PoE to be released.
So then Deadfire came along, and some changes were made and more risks absorbed. Marketing, if it was indeed absent (I just can't comment precisely, but my memories are of reading about this game as a side note on gamer press stories about Fig or crowdfunding, and then getting updates from Fig), is hugely important. Josh Sawyer commented about the voiceover expense in time and money (which I would argue is an indicator that the assumption that it is now required needs to be rethought - but that has been discussed ). The image I most associate with Deadfire's marketing is the picture one still sees on the Steam library page, of Serafen, Edér, Pallegina, and Aloth fighting off... pirate zombies? while tentacles attack? and maybe other pirate zombies are helping? or not? on a ship? It's dramatic and evocative, but evokes to my mind a Johnny Depp movie more than a dark, nuanced, thoughtful RPG meditation on the nature of souls, so I think there's some merit to the argument that it felt like a big departure from the first game.
Is familiarity a decisive factor in a sequel's sales? Not necessarily. But it doesn't have to be. It can be one of several non-decisive factors that sinks the ship (you see? I can employ nautical themes, too). For me it was not, because of my experience being skeptical about the first game's themes and then discovering that my skepticism was unnecessary and the seemingly jarring collection of themes meshed together well under the guidance of skilled writers. I started Deadfire and saw a significantly different tone, but still the same deep writing, the nuanced take on personal and collective motivations, and more rumination on the nature of divinity and the interesting notions of souls, reincarnation, and the Wheel. But those are the things I care about. Another player, who cares more about the dark feel of PoE, might have seen sun-drenched beaches and heard jolly pirate shanties being sung as jolly buccaneers freeboot from island to island, and looked away.
There are other factors. There are more streams and more let's-plays out there for people to watch before they buy. There are more isometric RPGs out there. Big megahits distort the market and players' expectations of what an RPG should be (Witcher 3, Skyrim, Original Sin 2 - and pour mettre mon propre grain de sel, I enjoy none of those three games). The state of localization was mentioned, an element I hadn't read about, and I agree - in such a narratively-driven game, it's important to get localization right if you're going to attempt it. None of these sound decisive to me, but I think each of them is a factor. And when one has an ambitious project (which Deadfire also was), those ambitions make the project more brittle.
Another earlier point that I think is very important: it's too early for anybody to say right now whether the game was successful or not. It sounds like there was a bit of "development hell" involved in the production of the game, at the least for Sawyer himself, so it's unlikely that he's going to be able to look back on things with perfectly clear eyes. For the sake of example: the game broke even recently. That's great news. Even if it never sells another copy, there is room for improvement. If I were an analyst for Obsidian, I'd at least do a rudimentary analysis of the project: if we could cut the "entire game gets VO" deliverable and only lose 5%, 10%, or 15% of sales, would the project meet our profitability targets for a viable project? There are likely other obvious questions to the people at Obsidian - I'm working only with what Sawyer shared in that presentation in Europe and my own speculations. But I wouldn't assume we know for sure that there cannot be a PoE3. I also caution that these discussions can't happen until more time has passed, so again: it's too early to say.
This was longer than I expected. I blame Boeroer and his love of Skyrim.
* I once described PoE by telling a friend that "it seems to have been a game designed around Voltaire's quote that 'if god did not exist, it would be necessary to invent him'." My friend, who has worked in the game development industry for decades, said "Such a high-minded concept usually isn't a good sign." We agreed that maybe the game worked because Obsidian didn't reveal that notion until well into the third act.