This past Wednesday, we spent an entire 8-hour day taking and answering questions about Kenya’s pursuit of al-Shabab in a live Q&A with our East Africa correspondent.
It was envisioned as a way to engage our audience around one of our top stories, and to tap into a clear need for reliable information about the issue. We answered questions on everything from al-Shabab’s funding to Kenya’s military equipment to casualty numbers, and ended up with a resource that will live on beyond the live chat as a valuable source of information.
We’ve done live chats before, but typically on a social media platform, and typically for a specific (and short) time period – our Asia team has hosted some successful Twitter chats, and our Learning English team regularly hosts live English lessons on Facebook. This was different because it was designed specifically to align with the way we saw our audience using stories on this topic, which led us to hold it on our website, run it all day, and go after deep, comprehensive responses to questions.
Using an analytics deep dive to design a social project
The story of Kenya’s incursion into Somalia was in the top 3 stories on our website every day for the 2 weeks after it happened. Having one item sustain interest like that was so unusual that my boss came to me and asked if we could do something social around the story, possibly a Facebook chat … at which point I had to admit something embarrassing; I hadn’t noticed the story’s popularity. Had I missed something? Was I not watching the social media sphere carefully enough? I put a post about it on our Facebook page to test the waters – the response was nothing above average.
To figure out what was going on, I dived into Google Analytics, creating an advanced segment for stories about Kenya, and looking at the time period since the military offensive began. First, I compared the advanced segment against the site as a whole and found that key numbers like pages/visit, time on site and bounce rate were all better for the Kenya stories than for the site on average. People were not only visiting these stories, they were engaged with them.
Looking more deeply into user behavior around the Kenya stories, I found Google and Google News (and its regional variants) were the most crucial traffic sources. Social media sites were almost non-existent as referral sources. Facebook and Twitter were quite low, and even social news sites like StumbleUpon weren’t driving much traffic at all. This was not a social story. People were coming in through Google News, reading the story, and then moving through our site to look at additional stories on the same subject – they were information-seeking.
In addition, perhaps unsurprisingly, East Africa and Kenya were making up a way higher percentage of traffic to these stories than to the site overall (and, interesting side note, a lot of it on mobile browsers, which led me to propose another project targeting mobile users that didn’t ultimately go forward).
So, the way we approached doing a live chat took those factors into account. We decided it needed to:
– Take place on our site rather than a social media site to capitalize on the Google News traffic around these stories
– Fulfill our audience’s desire for comprehensive and straightforward information
– Take place at a time convenient for East Africans
– Last over a long period of time, because people would be trickling in from referring sites rather than tuning in specifically for us
Putting Google Realtime to work
We solicited some advance questions on Facebook to get a sense of what people wanted to know. Our East Africa correspondent (the wonderful Gabe Joselow) drafted up responses to the most frequently asked questions, which he recorded as a short video and we then posted along with a transcript (and heavily hyperlinked to background material). At 9am Kenya time we began taking questions.
I produced the chat from Washington, managing the logistical side of it all – fielding the questions and editing, sub-editing and posting the answers. Our correspondent worked out of Nairobi researching and writing up the responses. We got exactly what we had hoped for in terms of the quality of questions, and more than we had hoped for in terms of the pace and number of questions (I had told the correspondent that this would slightly interrupt his normal work day, but we were both fully engaged on it for the entire 8 hours).
I also put Google Analytics Realtime to use to help track how we were doing as the chat was happening, keeping my analytics window open next to the chat window the whole time. I watched how the article was ranking on the site, and was prepared to take action to promote it whenever it dipped below #3 on the site (it didn’t drop until hour 7 though, at which point we were ready to let it peter out). I also watched refers very carefully to make sure they were lining up with my expectation of drawing traffic from Google News (they did – Google News #1, Google News Kenya #2), and to be ready to take advantage of any unusual refer sources.
I learned some stuff
I talk A LOT about the importance of having a strategy behind every project, and of understanding who it’s for and why it will work for them. But even so I usually come up with ideas based on an intrinsic understanding of the audience I interact with. It was interesting to start from the numbers and let them define the idea instead. The numbers supported my original hypothesis that this wasn’t a story that was happening on our social platforms, and also guided me to an understanding of where and how it was happening.
Oh, and I learned (once again) that it’s not fancy tools like CoveritLive or whatever that enable you to engage with your audience. This live chat ran on an article page using comments as the submission mechanism and a combination of replies and updates to the article text as the answer mechanism. Fancy tools don’t create engagement. Engaging with your audience gives you a reason to implement fancy tools once that engagement grows beyond what your existing systems can handle.