Feedback on EQ graphs
In addition to the comments I’ve received on my audience segmentation using EQ graphs post, I’ve also had some valuable feedback from a client I’m currently working with.
The project in question involves the redesign of a large information-rich website, using a collaborative approach whereby I am mentoring the client’s multidisciplinary team. I introduced them to the EQ graphs in order to help make sense of the qualitative research they had performed (in combination with a form of theme-based affinity diagramming) and to start to segment the website audience, with the goal of creating personas. They were interested in the technique, but it didn’t click with some of the team at first. After a bit of time to mull it over and talk about it among themselves, they got right into it and started tweaking the original idea.
Their tweaking included simplifying the number of attributes; our workshops came up with 15 attributes at first, which proved too many to deal with. Isolating six key attributes seemed to work the best. These were the attributes that really allowed the team to differentiate between the segments that were starting to appear, including “breadth of content required”, “level of scientific knowledge” and the ubiquitous “frequency of website visits”.
Another form of simplification came in the form of producing the graphs in Excel. At first the attributes were plotted on an electronic whiteboard which worked well in a group workshop environment. Doing this ensured plenty of discussion and debate, and once printed out, it was fairly easy to spread the pages over a large table and do some impromptu card sorting (a mini tip we discovered on the way is not to draw the vertical lines of the graph on the whiteboard—they’re a nightmare to ‘re-draw’ and make the graphs less legible). But during the whiteboard sessions, one team member had the bright idea of entering the attribute scores into Excel (sounds logical but I didn’t think of it!). This made the task of revising scores quite simple, not to mention the ability to generate nice neat graphs. The first generation of Excel graphs had the full compliment of attributes, but were soon simplified down to the six key attributes. Additionally a line of best fit was added (no it’s not mathematically correct since each attribute bears no relation to the next) as well as a moving average line, which in some cases helped us spot patterns.
I think the technique has worked well in this case because it catered for the mix of people involved. For those team members with a scientific brain (common place for this particular organisation), the graphs gave at least the illusion of science and hard analytical method, and they were happy with spotting trends and patterns as they would with scientific data. For team members who had a communications or marketing background, the graphs helped by adding to the general vibe (ala Dennis Denuto) they got from looking at the attributes and recalling the user research they were involved with.
One thing I have learnt from this application of the technique, is that it’s important to continue to main the mental link back to the users themselves. As with many analysis or ‘sensemaking’ techniques, it is quite easy to focus on the graphs and forget about what (and who) they represent. My catch-phrase during this work, which I’m happy to report the team started using themselves, was “take a step back…does this make sense?…is this what they were really like?”.
We’re now in the transition between segmentation and fleshing out the personas. No doubt I will soon hear that tell-tale sentence “but Bob wouldn’t do that!”.
Comments
Post a comment