Steam user reviews are the industry standard to estimate Steam sales. Depending on different things such as the release year, game size or genre the multiplier changes. To get a quick and easy rough estimate, for a current big SP AAA you have to multiply reviews by 25 (or 30 depending the case). Meaning the estimate for Starfield sales would be around 850K-1M on Steam.
Since on average users take a few days to make their reviews, this method works better waiting at leaset some days after the release, so that estimate is a bit too early, because real sales numbers may be a bit higher.
Starfield CCU all time peak was 269,177 users 15 hours ago, and since release almost every day has a new all time peak. It isn't true that the average player left the game. On average most players start to play a game and quit after a few hours, or buy it and don't play it. That's normal, nothing special about Starfield. Just look at the percentage of trophies in most games.
"Since almost everyday has a new all time peak." What are you talking about? There have been 3 peaks: early access launch, release, a day after release. That's not "everyday" having a new peak. Even then, it is the scale of the growth that matters, not peaking by 1 person. 7% growth from early access to launch, 15% from early access to day 2, I guess if this scales exponentially, in one week they'll have over 1m CCUs on Steam since it'll peak everyday!
But the crux is trying to downplay how bad the numbers are. If it isn't true the average player left the game, why the low growth? Because the low growth from early access could be from one of a few reasons:
1. Lots of EA purchasers refunded, meaning the EA CCUs therefore decreased, then lots of post-launch sales made that shortfall and account for the growth.
2. Nobody refunded, all the Starfield potential buyers purchased in EA period, with only a smaller amount of semi-interested parties, or those affected by word of mouth, purchasing accounting for the growth.
3. 100 million people bought the game, played 15m (all staggered as to not count in CCUs) agreed the game was good and decided to stop playing.
Option 1 seems the most likely considering very split perspective on the game and the Steam support spike. I would give it a maybe 66% chance of exceeding 300k CCUs on the weekend, but that is still barely over Skyrim, barely over Fallout 4. It's a regression. SteamDB estimates like 1-1.5m sold on Steam, based on Phil's CCUs comment, if Steam accounts for 1/4 of the players, that is 4-6m potential players (not sales because of Game Pass) which is half of Fallout 4's 12m day 1. And you know what is funny about the 6m number? It lines up with the estimations I had earlier when criticisng Peter Ovo who said the game had 20m players. 6m, maybe a majority playing for free on Game Pass, is again a regression.