Panel session at conferences are very uneven, and often they suck. Why? The primary blame can be laid at the feet of the moderators, who often don’t do enough to make the panels great. Charlie O’Donnell offers details in a great post:
Charlie O’Donnell, Why do panels suck and how can we make them better?
I spoke on a SXSW panel in 2011 that didn’t suck. I know it didn’t suck because the first person to ask a question told us that our panel was worth the whole price of admission to the conference and we got the same sentiment echoed on Twitter. The panel included myself, Emily Hickey, Ben Lerer, and Christine Herron and we spoke about startup mistakes.
The panel didn’t suck because it was engineered not to suck. Here are a few things we did:
- First and foremost, the panelists were carefully chosen. They aren’t the biggest VCs and entrepreneurs, but they’re some of the most thoughtful ones. Some of the most successful people simply haven’t tought much about why they got where they are—and even if they have, they’re just wrong about it because they’ve only scratched the surface. These panelists have seen both success and failure, and they’ve seen it from multiple perspectives—and on top of that, because I knew them well, I knew they’d be able to share those stories. Not everyone is a good storyteller, so choose carefully.
- The moderator had a sense of the story that should come out of the panel. I knew what I wanted to cover and what I wanted the audience to leave with. Panels are, or at least should be, stories, and a story is supposed to leave you with something. You should remember them because they make sense in a structure. Too many moderators pick something broad like “The Future of the Present” and ask vague questions like, “So what happens after now?” You’ll never get a tight story that people can leave with if that’s what you do. People either need to leave with a specific story or a sense of “If I believe x, this will happen, if I believe y, this will happen.” Moderating is hard and not everyone can do it. Respect the craft.
- The questions were discussed among the panel ahead of time. We vetted a bunch of topics and decided on the questions that would output the best answers. That also gave the panelists time to think about their answers. In fact, they were given a specific format by which they should structure their responses—to think about the tweets that we wanted to see before further explaining. So, when the question was “How can you tell what makes a good hire?” Someone would say a one line, tweetable, comment-worthy sentence as an answer before diving in further.
- Not everyone answered every question. Don’t you hate when they go through everyone in order and the last two panelsts basically say, “Yeah, what she said…” but they still take 5 minutes to say that. Some of the questions simply aren’t relevent to everyone. With our panel, each person was asked to answer only 2-3 of the questions, so the answers bounced around the four of us and no more than two people addressed any given question—unless they really had something ridiculously awesome to say.
- The panel talked amongst themselves. We disagreed on a few things, asked questions of each other. It was like we were real humans sitting next to each other discussing a topic. Amazing.
I think Charlie’s hit it on the head, but let me add a few thoughts.
Interviews are underutilized at conferences. In many cases, people who really don’t present well — despite having great ideas or being quite accomplished — are great when interviewed. Small panels — a moderator and two panelists, for example — can be great, especially when handled like a parallel interview by the moderator.
And panels should have no more than one person for every ten minutes: 50 minutes = four panelists max and one moderator, for example. That means in today’s fast twitch conferences, a 30 minute panel would/should/could have only two panelists and one moderator.
Last summer, I led a Future Of Work seminar series in five cities, and although we were allotting 45 minutes, I held the panelists to two in all cases but one, and that worked well. The time I had three panelists was a bit cramped, in comparison.