If the national popular vote mattered, then obviously in the general election the major candidates would have an incentive to campaign everywhere for every likely voter for their party.
To envision how they would conduct their campaigns we can look to the primaries where the nominees are selected and to commercial retail. In both cases—whether a person is selling a personality coupled with policy promises (the primaries) or a company is selling computer chips or potato chips (retail)—we know that the goal is close the sale to as many people as possible.
The first step is designing the product. A candidate, like a chip (hardware or chewy-ware, if that's an acceptable neologism), would have to suit the preferences of most people. To win a national popular vote a candidate would have to reflect the majority preferences on immigration (it's good); battling climate change (it's necessary); early child care subsidized by the government (important); better publicly funded infrastructure including high voltage power, high speed trains, and repaired roads (critical); limits on rounds per magazine in automatic weapons (of course); and a host of other topics.
Second, the candidate like the retailer has to build a brand. The means of branding would be to reach a national audience. That would lead to advertising on television content that huge audiences watch, like the Super Bowl or the Olympics. Currently 95% of all political advertising in the general election goes to local television in less than 10 states. We see in the critical early primaries, like Iowa, this same pattern within one state. But to build a national brand more than 5% would have to go to national advertising, such as on network shows. National branding would have to align with product-market fit; namely, the candidates' branding would have to appeal to the preferences of most people, instead of suiting niche audiences in a handful of states.
Third, just as Amazon delivers anything to anyone anywhere, candidates would deliver their messages to everyone. They would poll everyone in every town, which is feasible in the post-landline polling age we are now in. Their parties would offer to mail information, send out ballot applications virtually or by street mail, to everyone. The postal service would make more money; mail carriers would be part of the expansion of democracy to every precinct in the country. Social media advertising would go up in the aggregate, and would reach every demographic segment. It's important to note that the Internet does not care where you live. So using virtual mechanisms to reach everyone would certainly be part of national campaigning to win the national popular vote.
Fourth, television matters hugely, but in the current crabbed, confined system of competing only for swing votes in swing states, television advertising money in the general election goes to a handful of television stations. If the national vote mattered, middle and small sized television markets all over the country would get injections of political advertising.
Fifth, perhaps most interesting, local newspapers every four years would get a much needed injection of advertising. The cost of reaching their readers is relatively low and they offer a good way to present a candidacy. Newspapers in Mississippi and Missouri, North Dakota and North Carolina, and all the other areas currently in the land of ignored for the presidential candidates would not only get political advertising but they also would get interviews with candidates. They would have to hire reporters! That alone would reverse at least in part the sad trend of the last two decades of shrinking local news coverage.
Sixth, in big cities television advertising is too expensive. So in the top 10 media markets, social media would be used to reach voters in very large part.
Overall the amount of money spent would go up, but the amount spent per person would go down. This would be a relief for the badgered and beleaguered voters in the swing states who justifiably feel they are bothered way too much by advertising in presidential elections.