And in further 50/50 hindsight, PanSTARRS (one of the two biggest asteroid surveys currently going, also in Hawaii) dug through their archived data and found out they'd gotten in on 28 June and 7 July.
In each of those cases, and the ATLAS case, it was a short time span of observation. Long enough to show you that something was moving at a certain rate across the sky, but nothing terribly unusual about it... almost certainly just another bit of rock, probably out past Mars.
By the time SONEAR saw it, the object was about two million km away, call it five times the distance to the moon, and was moving along the sky fairly fast. They reported it as S511618 ("temporary designation" assigned by the discover.)
About ten hours later, the ASAS-SN folks saw it, now only about 1.3 million km away. After a pause (they don't usually track asteroids and had to look up the procedures for reporting one; they're variable star and supernova folk), they reported that they'd found a new object, with _their_ temporary designation assasn3. They observed it from both Texas and Hawaii, which was good; you get some parallax, which tells you how far away the object is. (We almost never get that; we have to rely on more indirect methods.)
Shortly after that, somebody noticed that you could link the observations for S511618 to those for assasn3. I was a little skeptical about this (not much data to work with, and it could have been a chance alignment), but if it _was_ a real link, it should be possible for somebody to observe it and get more data. Alerts went out to amateurs. I was hoping somebody in southern Europe or Africa would get it. (There are some observatories in South Africa that were ideally situated for this.)
And in fact, some observers in Italy and Ukraine _did_ get more data (not easy; the object was low in their skies.) But before then, ATLAS and PanSTARRS went back through their old data and realized they'd gotten the object days/weeks earlier.
This business of figuring out which objects are likely to hit and which aren't isn't easy. Sometimes, we get the data we need and it all lines up beautifully. Sometimes, we don't.
I should note that after it flew past us and was too close to the sun for any optical telescope to observe it, Arecibo got some echoes off it. That's helped to nail down the orbit. Haven't heard if we'll get images from it (which was the main reason Arecibo pinged it).