Making sense of thousands of stories, suggestions or complaints

Stories are fabulous things, but there’s one thing about them that makes them a real bugger to work with. They’re big, wordy and take time to read.  And if you’ve got loads of them (or indeed anything over a couple of dozen) it’s well-nigh impossible to see patterns in them or analyse them. Once I’d got past the initial hurdle of how to generate (or more accurately collect) stories in an organisation, this was the show-stopper in early Narrate projects.

So, when I was in Washington a little while ago and caught one of the new (and sadly not re-commissioned) TV series, I had to laugh at the ease with which they brushed past “the difficult bit.” The show was FlashForward and revolves around everyone on the planet having a vision of themselves at a set point in the future. And then the fun starts… The maverick investigators (naturally) then build a website called Mosaic – into which everyone can enter the stories of their visions, and then the investigators can pick up the patterns and matches in these millions of stories.

If all you have are raw stories, that’s a near impossibility. There are technology companies that will tell you they can do semantic or textual analysis – but they are deeply flawed at best. There are human analysts who can read a sample and tell you what the teller meant, but as I found out when I did that early in Narrate’s evolution, that only gives you the analyst’s interpretation of the individual story (often very different from the teller’s meaning) and you get any number of cognitive biases kicking in, so that you miss stories that show you important but rare information, you get blinded by what you expect to see, etc, etc.

Thankfully, we’re now past that point. In the past two years, we’ve run a number of major projects using Cognitive Edge’s SenseMaker™* software – collecting raw stories (in their badly spelled, ungrammatical, sometimes garbled or terse forms) but with the tellers adding meaning to their stories by showing us what they see in their stories. They use shapes to show us the meaning like these triads (taken from two recent projects, one was to understand sales issues in a multinational, the other was the international pilot for a proposed global cultural/anthropological collection of stories from all cultures):

People just mark the triangles to indicate how they see what was going on in the story. Simple and quick.

It means we keep the teller’s original meaning attached to the story, but we also get something even more useful – a way of analysing the mass of stories. Because when you have collected lots of stories, you can look at a single triangle and see where they’re all placed. The triads below are the results from the triads above, with each story represented by a dot.

(The triad above, Customer emotions, has almost 1200 stories in it (one per dot), but we can see where the stories are and see where the main beliefs fall. The triad, Justice below, also has the dots colour-coded to show what stories people thought were positive and what stories they thought negative.)

So finally we get to see the overall patterns from thousands of stories (or customer reports or suggestions or any other form of qualitative input) – and see how the original tellers see the world. But throughout, we’ve got the original stories tucked away beneath each dot, so we can, for instance, look at all the stories in that little cluster in the centre of the triad on the right. There are twenty:

You can see on the left of the box all of the titles of the twenty stories – including multiple ones about the same thing. And on the right of the box you can see the text for the story that’s been selected – “El Oso Sonador”

So, finally, we have a tool that is exceptional at showing us the patterns that emerge from masses of qualitative data like stories. And then allows us to zoom in on particular areas of interest to see what stories or material drive that area. Given how easy and cheap it is now to generate stories in volume (via all sorts of collection techniques – worth a later blog on its own), this then becomes great at seeing the overall picture and sifting the important and useful material out of the mass.

For anyone gathering large amounts of material from audiences – be they customers, employees, residents – it’s invaluable to add a little extra on the input side, but be able to do fast and accurate analysis on the back end.

*By way of full disclosure – we’re part of the Cognitive Edge network of practitioners and therefore licensed to use and sell SenseMaker™.