Evaluating Titanfall and its first month sales essentially boils down to what lens you're observing them under.
1. From a pure "new IP and nothing else" POV, it sold absolutely incredible. If you factor in all of the marketing costs (and I mean all of them; website hype, commercials, etc.) it still sold very well as a piece of software.
2. From a Call of Duty killer standpoint, it sold under what it should have, but it was kind of put into a position to sell under what it should have, given its console-exclusivity. I'd say it was also set up to fail from a software standpoint given the lack of single player campaign. But still, it didn't have to immediately kill of COD on its first go-round (an impossible task), and the fact is it still sold quite well.
3. As a system seller? It definitely was one, but it's quite difficult to discern how much of one it was. For the month of its release, the XB1 coming in second in the NPD charts is seemingly damning, given the hype and marketing and whatnot. However, is it arguable that numerous XB1 owners who bought the system in the months prior were buying it primarily for Titanfall? Definitely. Which leads us to...
4. Titanfall's performance as a killer app. Given those NPD numbers and the money Microsoft has sunk into it, Titanfall's performance has to be deemed disappointing - at least in terms of pushing XB1's. Is that the game's fault? Perhaps, perhaps not. Perhaps Halo 5 couldn't have pushed the system into first place last month. Who knows. But in terms of what Microsoft was expecting out of the game in terms of pushing systems, it failed. And I think that's where you're getting people saying "oh, this new IP that's online-only only sold a bit over a million in four weeks, what shit!" In terms of raw numbers, the game did pretty much all it could on the XB1's userbase. The 360 numbers were probably hampered due to the two week delay (and perceived inferiority), and who knows how it did on PC.
So how did Titanfall perform? Terrific and terrible. Simple as that.