Now I’m back from New Zealand (boo! hiss!), we have a new website (yay! phew!) and Andrew is inbound to take over the day-to-day evolution of the site (hallelujah!) I’ve been taking a bit of time to revisit some other ideas that have been rattling around in my head the last couple of years sparked by the opportunities and challenges the new site was likely to offer.
So potentially this is the first of three or four posts about what I might try to tackle next.
One thing I have long been interested in is finding a way of making web analytics data both more available and more understandable to the people actually providing the statistics, analysis and commentary we publish. Whilst analytics tools like Google’s Universal Analytics have become increasingly powerful they have also become increasingly complex. They have also developed a vocabulary that is focused on e-commerce and advertising — the language of conversions, clicks and eyeballs.
At the moment ‘raw’ access to the analytics can often be as harmful as helpful as it is easy to misinterpret the data and focus on the wrong metrics.
It has become clear that this is far from a local problem. A couple of blogposts this week drew my attention to the Ophan analytics platform at the Guardian while a separate article introduced me to Lantern, a project with similar goals at the FT. I also found articles about not dissimilar undertakings at NPR and the New York Times.
They are all trying to solve the same user needs — to develop an analytics service that speaks the language of their core users (journalists) and just provides the metrics that matter to them.
As Renée Kaplan at the FT says they are trying to;
“..translate (analytics) into journalese, into newsroom vernacular, all of the metrics of engagement, all of the metrics that show not only how a story is doing in terms of the more traditional pageview performance, but also how our audience is integrating, reacting, and interacting with it..”
While the Guardian and FT built new platforms the approach at both NPR and the NY Times has been more about translating the metrics Google (and others) already provide in to using more palatable and actionable for their audience;
“NPR doesn’t introduce any new form of measuring analytics with its dashboard; rather, it takes existing information and presents it in a way that’s more easily digestible.”
It seems to me there are a few steps to making this sort of thing work.
Identify (via user research) what metrics are actually helpful to the teams providing the content. Some things that spring to my mind could be;
- How many people viewed this page?
- How much did they read?
- Who read it?
- Did they download anything?
- How did they get to the page?
- Search? Social media? Browse?
- If it was search what were the search terms?
- Where did they go after the page?
- Has it been shared on social media? How often?
Then you need to find out the best way (if at all possible) to actually capture the data for these needs. Now that might be using the API of an existing product (Google Analytics, Chartbeat, Parsely etc) or building on top of something like Splunk to spin your own tool out or even start from scratch Guardian style. Whatever works best.
Finally find a way to present it that is genuinely user friendly, uses the language of its audience and is available via the right channels. One thing I liked about the NPR solution is that it automatically pushed emails with the top level metrics to the owners of the content.
I also always loved the idea of the /info/ pages on GOV.UK and I think there is potentially something interesting possible about offering metrics that follow the same structure as the site.
Browse to https://www.ons.gov.uk/economy (maybe just adding /info GDS style) and get high level metrics for the Economy section.
Go to https://www.ons.gov.uk/economy/grossdomesticproductgdp and get everything related to GDP but go to Second Estimate of GDP: Quarter 4 (Oct to Dec) 2015 and you get detailed page level metrics.
Clearly we are not a publisher on the scale of any of these organisations (not even GOV.UK) but I think the parallels are valid and while we aren’t fighting for subscribers or advertisers we do have pressures as well as limited resources so the ability to make more ‘data-informed’ decisions seems worth the investigation and potentially the investment.
To do this right you would probably need to find a way of folding in social media analytics from Twitter, Facebook and LinkedIn (that is pretty much where we focus our efforts) as well as email newsletters and the like.
Sounds like a lot of fun to me.