The only way to even *know* what readers might like is to allow them to read and share on the open Internet.
Knowing what it's like inside a media company, let me state up front that there are epistemological problems in deciding from the outside why The Daily failed. I've yet to see an article about The Atlantic that understands how our site works under the hood, and why that has led to a relatively successful few years.
That said, I do have a few thoughts about The Daily that I'd like to offer alongside my colleague Derek Thompson's.
1. The Daily was built on the false premise of a "general reader." When I hear general reader, I think that a media executive is imagining himself and his friends (you know, normal guys) and intending to produce a bundle of content for that hyperspecific DC-to-Boston-went-to-a-good-college-polo-shirts-and-grilling demographic. As a result, anything that falls outside the boundaries of the interests of this presumed Joe Six Pack will be deemed too "cool, quirky, nerdy, obsessive or snarky." That's not to say that such a publication will never run such pieces. My old pal Ben Carlson wrote quite a few. (So did Zach Baron and Sarah Weinman.) But you're fighting the institutional gravity. And you'll have to build defenses into the story that are the writerly equivalent of "I don't meant to nerd out here, but..." Because otherwise the bros running the grill will throw you in the pool.
This is not to say that media properties cannot be built with the goal of reaching the mainstream, if by mainstream we mean very large audiences. This, very clearly, can be done. (See: HuffPo, Gawker, etc.) But! And this is a big but! These sites have been built up like sedimentary rock from a bunch of smaller microaudiences. Layers of audience stack on top one another to reach high up the trafficometer. The various voices of their bloggers attract layers of readers. New York attracts a different layer of readers. Left-wingish attracts yet another. Their big investigative pieces add more. And their super niche pieces sometimes explode -- say, Matt Buchanan's tech truth bomb -- precisely because they originate inside a niche. It doesn't feel like your uncle from Evanston is telling you the latest thing about iPhones or queer dance or SEC football.
2. The Daily was built on the false premise it could control the distribution outlet. Being in media is terrifying right now. Whereas in the old days, you wrote something and then a fleet of people printed it and handed it to X hundred thousand people so they would read it, now, the fleet is gone. You are alone out there in the ocean and there's not much that anyone can do for any given story to make sure that people read it. Seriously, since the fall of '08 vintage Digg, there's not much anyone on your Internet's favorite websites can do aside from stick a story on the homepage, tweet/Facebook/tumble/Reddit/LinkedIn it and then pray. We do not control the distribution of our work. Period. It's horrible and bizarre and it is also the way that the media world works now. You can't push; the content has to pull. (Especially if we are talking numbers in the millions.)
3. Which brings us to the cardinal sin, The Daily was not tuned for sharing. Obviously: it was mostly locked up on the iPad behind a paywall. Less obviously: how would anyone build an audience growth strategy without relying on huge social hits and a distinctive voice? How many content sites or apps have you visited as a result of marketing? Then how many did you go back to? Without the actual mechanism of HUMANS DROPPING A PAPER AT YOUR DOOR, there's just no way to keep people coming back without A) a voice/POV/knowledge base that is impossible to find elsewhere B) massive traction day after day after day after day after day after day after day in the social world, light and dark forms alike. I haven't seen a digital media play succeed in the last five years without both of these factors.
It's not a technical proposition to get people to share your content. Sure, there are things you can do that marginally improve your share rates by placing the buttons in the right places, etc. But the main determinant of social sharing is the quality, tone, and form of your stories. The only way to figure out what works -- because it's constantly evolving -- is to keep sending work you love out into the ecosystem and seeing what gets amplified. Then, you take that feedback, write another thing you love and send it back into the field. If a bunch of stuff sits behind a paywall, that iterative cycle gets broken. You don't know what's working.
Let me give you a really short example from our work. There's a classic longform convention in profiles of people. The writers tend to drop in to the story sitting with their subjects. They describe what they look like and provide some color about the situation: are they eating? how'd they get there? was the publicist a stickler? does the person appear to be on drugs? Sometimes we get a very short quote from the profilee that is indicative of the person's affect and intellect. Then, the story, by which I mean the action, really begins. In particular, the stakes are explained lower down in the story, several grafs in. These description-rich ledes come first because that is how it is done (and it can be artful as hell when done perfectly).
Well, we found this sort of thing bombs for us over and over. Maybe we don't profile the right people (disagree!). Maybe we're terrible at writing these sorts of ledes (perhaps!). Or maybe, just maybe, the form doesn't work very well to capture the attention of people who are clicking through from an email, IM, Facebook, Twitter, Tumblr, or Reddit. You just don't know what's happening for several hundred words in some cases. And before the writer answers why you're reading, that person is gone.
So, we switched it up. Rather than say, "Well, long, narrative pieces of writing suck for us," we changed the ordering of our narratives. We get the stakes up very, very high. I'm talking within the first 100 or 200 words, even if the story is two or three or four thousand words. And if we want to keep that narrative lede and we know we have nuggets down low, we drop in a tl;dr box to capture the kind of reader who wants to know what the point is exactly. (If you're a regular reader, you will start to notice this pattern.)
Now, this strategy might not work for other people. Hell, it might not keep working for us. But we have a high hit rate with the pieces that we put extra time and special effort into. And that's the feedback loop you want; that's the one that makes us happy. But the only way to even *know* what readers might like is to allow them to read and share those pieces on the open Internet.