Organizing at Scale

From DevSummit
Jump to: navigation, search

facilitated by Matt Holland

Matt used to work at True Majority - around a million on list. Now he's at Avaaz - 43 million Some principles are the same no matter how big your list is - clickthroughs, action rates, etc.

Pace of list-growth

When Matt was at True Majority, they started with very steep list growth. They predicted a very over-optmistic growth over next five years. The curve always eventually gets flatter, but how long that takes varies from org to org.

Avaaz - launches campaigns based on what's happening in the world, specific opportunities in the moment.


The Far Side of 40 million

  • Avaaz - 43 million email addresses
  • 5 million as of 4 years ago, when Matt started
  • Has also managed lists of 750K
  • Even asking a question of a database can take days with 40 million+ records

Credible Actions

Big rule at Avaaz: there has to be a credible theory of change, not just petitions for the purpose of growing lists. People are smarter than we give them credit for; they'll participate if we show them that their participation can actually make a difference.

That also includes telling people to forward to their friends -- show them a credible reason why sharing with their friends will make a difference.

Testing

Pay careful attention to the wording -- you might test a message many times. Response rate goes up or down.

Question about advertising: almost all word-of-mouth. Avaaz occasionally does advertising when it's something that's really hot right now.

  • Testing is not just about A vs. B
  • Also, exceeding a certain threshold of engagement
    • e.g., If this email doesn't convince 5% of people who open it to take action, we're not sending it (beyond the test list)

List Growth Optimization

Pay attention to sudden changes in list, and turn up the volume on anything that works.

  • Example: a campaign targeted to Brazil got a spike from people sharing on Orkut, so they added an Orkut share button.

Sasha: Messages that target people emotionally get new list subscribers, but it's difficult to find ways to deepen people's level of participation.

  • Matt: it might not always be a matter of getting people more involved. Sometimes it's finding the people who are going to get more involved anyway, and giving them something to do.

Ask people why they care. They give you important feedback. Asked people to write essays about why they wanted to be brought to an Avaaz event. Got great content as a result.

  • 350.org approach - much more specific in tracking people's behavior. e.g., this person signs up for a lot of events but then doesn't come to them.

Knowing when outreach doesn't work---when you're perceived as taking advantage of something in the news.

Emotion in emails: you can't fake it! It has to be real. The emails that were "what I really think" were the most successful ones.

The year of Copenhagen, Avaaz made a decision not to focus on list growth at all, only focusing on action at Copenhagen. They decided that that was the wrong decision; they needed to make the list-growth focus continuous, regardless of where they were directing action.

If you get to the point where you're the size of change.org or Avaaz and you decide to build your own CRM, it gets a lot more difficult. Security becomes your problem. Email reputation becomes your problem. (Send emails all at once = get on spam lists) Whole bundle of problems that don't exist when you're on an off-the-shelf CRM.

  • But the benefits of running your own CRM: you control the product. You can add an Orkut button if you get a lot of uptake in Brazil. You can run better experiments (e.g., testing different amounts of money to ask for). Customize for different currencies.

Meorio - has a good custom system. Grassroots community building.

Question: What data do you need to know about your constituents and why?

  • Collecting the history of every interaction with the org -- eventually becomes more than you can possibly use. Becomes a barrier to research, rather than an aid.

How long to run an AB test for? Avaaz worked out a formula for knowing when the difference between results in an AB test is statistically significant.

Avaaz doesn't change campaigns a lot from country to country or vertical to vertical. Underlying principal of the organization: people are people.

But you do need to slow down emails to people who never open them, or spam filters will ding you.

Question: this discussion is so focused on email -- are we missing the right medium to get to some people?