This post originally ran in OnlineBehavior.com in January 2014.
It’s that “most wonderful time of the year” when annual reports are generated explaining what the digital team did and the value it produced. Yet, if like many companies, you report value as page view volume, visit volume, social likes or tweets, then your boss is most likely to say “So what?” And if they don’t then they should be. For example, there’s no way to tell whether an increase of time spent on site is confusion or engagement. But these are topics for another post.
If you are lucky he or she will grudgingly renew your budget for another year on the simple idea that digital marketing is not going away any time soon. But there is so much more you could be doing with essentially the same data.
Welcome to the miracle of audience segmentation. Yes, it is true that there are actually groups of people with similar goals who come to your sites, mobile apps and interact with you in the social sphere. But when your digital marketing metrics are reported as “most popular blog post” or “most liked page” or even “most popular video,” even though it sounds like valuable information about what people are doing it’s really not.
What matters is the discrete audience groups behind those KPIs and to find them you have to separate the signal from the noise.
A good example comes from the non-profit space. This is the busiest time of the year for organizations that count on donors to meet their missions (and if you have not yet picked a charity to donate to, then use this link to find a good one). As with many companies, in the non-profit space there are news releases, social postings and tweets, site updates, and numerous ways to “engage” or encourage the desired action, in this case donation. Basic KPI reporting for charities would likely include dollars donated, top pages, and maybe even list marketing campaign or newsletter click-throughs.
Yet this completely misses the point because nonprofit data is filled with noise. They are commonly inundated with job seekers, so the audience must be grouped before the metrics have any meaning.
Grouping the audience means creating mutually exclusive definitions which capture 95% or more of your existing non-bounce visits. So, we might start with a 3 month time period to get a reasonable sample of diverse audience visit behavior. Then we might create a handful of segments that group behavior.
Anyone who landed on the jobs page or consumed it within the first 2 pages of their visit goes in the “Job Seekers” bucket. Anyone who immediately goes to the event space section becomes “Event Planners” and those who shop the holiday “give in-honor-of section” are the “Donors”.
Once the initial buckets are defined – and mutually exclude each other by definition – then look at what’s left to see what was missed. Should an existing bucket have an expanded definition? Or perhaps there should there be a new audience type? When you are finished there should be about 5-7 groups. Now take those same KPIs and run them through audience types and your report sounds more human.
- Job seekers really like the photo contest and content about volunteering, but our Donors prefer news content specifically about Syria.
- Our paid search campaign on these terms is generating less donation interest – even over multiple visits – than these other terms.
- While we are seeing more products purchased this year overall, most purchases are coming from existing donors. Only 5% of our newly recruited donor targets are converting, many of whom we acquired in paid search but 75% simply bounced off the site.
The best part of grouping your audience is that you’ve now become a storyteller, not a KPI data generator, and everyone loves a good story.
If you would like to learn more about data storytelling, check out the Digital Analytics Association Seattle upcoming 2014 speaker series. If you’re looking for help executing a measurement framework, give me a call.