Today I attended a fabulous session of the Web Managers Roundtable at AARP which featured a presentation by Dan Siroker, the former Director of Analytics for the Obama campaign, and co-founder of the A/B testing company Optimizely.
As the group of 50 or so professionals were presenting themselves, I couldn’t help but notice how for so many organizations analytics falls in the same category as social media. Then again, as Dan mentioned in his presentation, on the Obama campaign anything that was not well understood went under the New Media umbrella 🙂
Titles aside, clearly the Obama’s campaign did many things right as demonstrated by Dan with a couple of data points:
- Facebook friends – 2.4 million for Obama vs 0.6 million of McCain
- YouTube video views – ~100 million for Obama vs ~20 million for McCain
- Unique website visitors – ~130 million for Obama vs ~30 million for McCain
- ~$500 million raised online
In his presentation, subtitled “Lessons from Obama to Haiti”, Dan Siroker who had been approached also by the Clinton Bush Fund for Haiti in helping them optimize their fund-raising campaign, shared five lessons:
1. Define Success
Define quantifiable success metrics, for example:
Website click thru rate = # of clicks / # of impressions
Email signup rate = # of signups / # of pageviews
Raised money per recipient = $ amount raised / # of recipients
2. Question Assumptions
This was one of the most fun parts of the presentations because Dan engaged the audience in a live multivariate testing. He showed us several variables — for the media and for the button on the splash page of the Obama campaign website, that were considered two nights before the Iowa primaries — and had us all vote for what we felt would be best. Very few of us guessed what the data had shown to work best — the “Learn More” button and a “Obama family photo”. And that was exactly the point — that by questioning all assumptions and relying on data, you can arrive at gradual improvements that lead to real results.
Selecting the “Learn More” button for the email signup over the previous “Join Us Now” button had increased signup rate by 18.6%.
Choosing the “Obama family photo” for the media choice on the splash page had carried additional 13.3% of improvement over.
The majority of the participants, myself included, had chosen a rousing, inspiring video from the Springfield conference but as Dan explained, the function of a splash page is to quickly skip it and get into the site, thus the media choice had to be something simple — a long video had its place but on the splash page.
The combination of these two optimization factors lead to a 40.6% combined improvement which lead to approximately 2.88 million additional email subscriptions, 288,000 additional volunteers and $57 million additional contributions. That is real money!
3. Divide & Conquer
This lesson was illustrated by another demonstration of a multivariate testing, this time indicating that audience matters — what might work for one audience, or time, or place, might not work for another audience.
4. Take Advantage of Circumstances
You can never predict life, so with all the data and powerful tools you might have at your disposal, you still need to have the flexibility to adjust course depending on what circumstances present themselves to you. A couple of well known examples illustrated well this point. Being present in the moment is important for all of us but probably even more valuable to a number person!