Here again, we don't fully understand the problem because we can't model all of the inputs. I think generally that hype is less effective than those who urge its use believe it to be, for essentially the same reasons that lineup selection is less important than people assume it to be.
When Walter Cronkite was broadcasting, the first words out of his mouth would be the essential facts of the day's top news story. When a modern local announcer is broadcasting, he or she will "tease" the audience repeatedly. He'll come on 10 minutes before the newscast with a ten-word bulletin, "THREE DEAD IN A HOUSE FIRE IN OLATHE!; DETAILS AT FIVE". Then, at 5:00, he'll come on and say, "A house fire in Olathe kills three and sends another to the hospital in critical condition; we'll have all the news following this commercial." "This commercial" turns out to be a 3-minute sequence of 10 to 15 twenty-second commercials, then the newsman returns and says, "Electrical wiring is suspected as the cause of a house fire in Olathe which has killed three people and sent a 70-year-old relative to a critical care unit. Randy Reporter is standing by, and he'll have all of the details on that story in just a moment, but first. . ." and then he goes to a different story, and a different story, and a different story, each one of them a mix of a few facts and some hype for another story to be told later in the broadcast. After four minutes of this, they'll do a round-the-table; the weather guy will say 20 words, the sports anchor will say 20 words, and co-host will make a joke, and then they'll do another 3 minutes of commercials.
They THINK that they are "building the audience"; they are building up the audience by telling everybody just a little bit and a little bit and a little bit about what is GOING to happen later in the broadcast. But by delaying, and delaying, and delaying, they are also sending a sub-concious message to the audience: None of this is really important, and we don't actually care anything about it. If it WAS important, we would tell you about it NOW. But it isn't really that important, so we'll tell you about it later.
So we don't watch the news anymore; we just steal headlines from the television and go check the facts on the internet if the story seems relevant to us. Would the old Walter Cronkite approach still work, in the modern world? Who knows? It would work for me, but then, I'm 66 years old; I'm not the target audience.
I am trying to get to a point, which is something like this: that playing the percentages doesn't actually work. The percentage advantage that you can SEE is usually undermined by some unseen disadvantage. This is why lineup selection has little impact on actual runs scored; it is why Hype doesn't actually build audience share. It is why negative advertising doesn't actually work; I am never going to vote for Donald Trump, but the Democratic convention is sure trying like hell to MAKE me vote for him. Every time they accuse Trump of being a bully, I feel a little twinge like I ought to vote for him, just to reject their stupid name-calling campaign. Somewhere in a backroom is a guy with a study who has found that people are 17% less likely to vote for Donald Trump if we call him a bully 22,000 times. It doesn't actually work.