Wins and Losses should be the primary metric in a system with relative parity where all teams play the majority of other teams.
The Big Ten, for instance. There's relative parity within the conference, and each team plays the others at least once, and sometimes twice. There's some variance in schedule difficulty if you get multiple of the best teams twice and another team gets the lowest teams twice... but it's not a huge difference. Wins/losses determine ranking within the conference.
Now look at the MEAC. Same story, when looking within the conference games themselves.
But is a team that's 9-5 in the MEAC better than a team that's 8-6 in the Big Ten? No.
You need to rank the teams in another way.
If wins/losses aren't really representative of quality across a field of teams so widely varied in strength, you need to look at a different metric.
Enter efficiency. Kenpom and Barttorvik both base their rankings primarily on efficiency, making some adjustments, and so does NET.
If your DEff is usually 93.0, you'd be expected to fare even better than that against a team whose OEff is less than 93.0 already, but worse against a team whose OEff is 105.0... which leads to adjusted efficiency numbers. How good is your efficiency when taking into account your opponent's average efficiency in their other games?
We played good defense against Stonehill but truly, truly awful offense.... and that resulted in a 1-point win against a bad team. Our OEff was 84.7 that game.... but when taking into account the poor level of Stonehill's DEff (116.2), that number gets adjusted down to 80.3
That poor performance is part of our record, and it informs our overall ranking. If we'd beaten Stonehill by 40 (like we beat similarly-ranked Columbia last year), that would also be part of our record.