https://www.theatlantic.com/politics/archive/2017/09/campaigns-direct-mail-zero-effect/541485/https://www.vox.com/policy-and-politics/2017/9/28/16367580/campaigning-doesnt-work-general-election-study-kalla-broockman (Vox article is more detailed)A few notes:
1. The reason behind this they postulated is that partisanship/tribal identity makes persuasion extremely hard. They did find that persuasion is still much more effective in highly-personal, substantive discussions, but that is not too practical in large scale campaigns. The gist is that a lot of people are persuadable, just not by these "drive by" campaign tactics, like advertisements, mailers and brief canvassing.
2. They found that primaries were far more receptive to campaign tactics, largely because the wall of partisanship is no longer there, as it all occurs within their own party. It's much easier for a fellow Republican to convince another Republican, than it is for a Democrat to convince a Republican. It's not even close. They also found persuasion effective with ballot initiatives, mostly for the same reason.
3. GOTV and turnout operations are still effective. In this sense, a good campaign strategy might be to treat each election as a base turnout election. You do what you can to persuade people, but invest far more resources into turning out your reliable voters. Spending tens of millions on advertisements months before the election is largely a waste of money.
4. It was a very comprehensive study, involving a lot of other studies. They used a lot of data from campaigns, where they reached an agreement to share for academic research like this.