Net Success Rate is a pretty simple; do they get 50% of the yards to a first down on 1st (generally 5 yards), 70% of the yards on 2nd down, and 100% of the yards to gain for 3rd and 4th down. It does not punish turnovers and does not reward big plays or touchdowns. It's really a measure of what percent of plays are you "on schedule." There might be a small bit about EPA (estimated points added per play) that the play must also be greater than 0 (increases expectation of win per play) on Parker Fleming's model but that is just a small change to the overall number.
So, for example, throwing 10 straight plays of 6 yards to get a touchdown and a single 60 yard touchdown would both give you a value of 1.0. But then on the next drive if you go three and out with no yards gained then punt. The longer drive would give you a value of 0.77 while the one big play gives you 0.25. Even if you went 9 straight plays of 6 yards and then had a turnover, the overall Net Success Rate would be higher with the 13 plays (0.69) than the 4 plays (0.25).
The model is generally successful in reflecting how the game went but you look at USF's 34-7 victory over Boise State as the least impressive win on a per play basis and you can see where this model has its flaws. USF had a bunch of nothing drives and then big touchdown plays where Boise State had consistent drives that ended in eight drives ending in USF territory (4 on downs, 3 on fumbles, and 1 to end the game).