Pollsters often use the respondent's answer whether they're likely to vote as a factor in including them in the survey. Usually more people say they'll vote/think they'll vote than will actually vote. I've talked to people on election day who tell me they'll vote and they don't end up voting. Are Democrats going to convince people who have decided not to vote to go to the ballot box? What will they say or do in the last three weeks they haven't been doing for the rest of the cycle?
The Democrats have let it be known that they're spending a lot on turnout, with a lot of paid operatives and that they have a lot of data. No matter how many operatives you have or how much you know about someone you aren't going to make them be home when you knock or answer the phone when you call. Anyone who has done the ground game knows that the vast majority of people aren't NH or you get VM.
I know that having information on voters can help but it'll help more with party faithful, who are high turn-out voters anyway. Knowing that someone is pro-choice is great but anyone who's been out there knows that most people you do get at home don't want to chat with you. If they don't, it's smart to just move on.
Democrats may win the ground game. Or they might not. In San Diego they twice as many walkers as we did and walked Republican neighborhoods while we didn't walk Democratic neighborhoods. We still beat them by more than people expected. I don't know that the Democratic ground game isn't accounted for in polls, but if it isn't I'm skeptical it'll mean enough to make up for the deficits we're seeing.