Fresh off the CommsNetwork mailing list comes a discussion about how to best encourage an audience to ask questions. It’s a common problem, and there probably is no simple soluation – still, there are lots of good ideas. Here’s my summary:

Various forms of “planting” questions

• Plant 1-3 pre-defined questions in the audience beforehand
• Encourage specific people you know to prepare a question of their own choosing
• Put pre-formulated questions or topics on certain seats as encouragement

Collect questions off-line

• Prepare a wall for post-it questions to stick on (both gives passive questions and lowers the threshold of a topic being brought up)
• Put blank question cards on the seats, collect them during the event
• Generate questions per table over lunch
• Collect questions beforehand, draw prizes among submitters
• Contact “High Potentials” for questions beforehand
• “Snowball” questions: Have the audience write them on paper, crumple and throw them to the front
• Make sure the first question asked is a significant one – one that management can answer in a way to demonstrate openness for frank discussions – as it sets the baseline for others

Small Group Interaction

• Localize the call for questions or group the audience around tables
• Have a facilitator or a familiar name to generate / ask questions
• Sit the speakers amongst the audience to encourage micro-feedback
• Make short breaks and have managers wander around actively asking questions

Large Group Interaction

• Encourage questions during (as opposed to after) presentations
• Have presenters ask simple, non-scary questions to the audience (“How many of you…” – followed by question to someone who raises (or doesn’t raise) his hand)
• Distribute small prizes (car wash coupons, movie tickets) for the first people who ask a question

Meeting Technology

• Use a question submission technology like Crystal Interactive
• Use interactive voting systems

Some time ago, I stumbled across an crude yet interesting way to measure the internet reputation of any “brand” – be it companies, products, or people. The concept is simple: Count the number of times somebody says something nice about the brand, and compare it to the times somebody says something not so nice about it.

• As a counter, we use Google
• As measure of “nice things”, we search for mentions of (“I like XYZ” OR “I love XYZ”)
• As measure of “not so nice things”, we search for mentions of (“I don’t like XYZ” OR “I hate XYZ”)

Tabulated against each other, “number of nice things” vs. “number of not so nice things” gives you an impression of the overall online reputation. To relativize things, we can additionally introduce a measure of how often XYZ is mentioned at all.

This measure is by no means perfect or even remotely scientific. Search results could be distorted by all sorts of effects, and the reputation search criteria could be much more refined. Still, it’s a wonderfully simple instrument which might be useful to measure what the online world thinks about a brand. Try it!

And just for fun, some online reputation ratios:

• Paris Hilton: 32′300 positive vs 20′600 negative (ratio = 1.57)
• George W. Bush: 13′300 positive vs. 11′200 negative (ratio = 1.19)
• Microsoft: 58′400 positive vs. 103′000 negative (ratio = 0.57)
• Apple: 295′000 positive vs.  66′400 negative (ratio = 4.44)

Disclaimer: If you don’t like math, you might want to skip this post.

I have been asked if I had an idea for a numerical indicator showing the amount of diversity within a group. Let me make an example:

We have a population of N=100. Each member of that population has a favorite color. The population is therefore divideded into subgroups with the size of n(i) representing those favorite colors (e.g. n[blue]=50 people, n[red] = 20, n[green]=10, n[orange]=10, n[yellow]=8, n[purple]=2).

What I’m looking for is a one-dimensional, linear representation of the diversity of color preference, limited to the extremes 0 (=no diversity, i.e. everyone has the same favorite color) and 1 (=full diversity, i.e. everyone has a different color).

Specification: The difference between two colors can either be measured (a) nominally (i.e. light yellow and dark yellow are equally different as green and red), or (b) on an interval scale (i.e. red and green are measurably more different than light yellow and dark yellow). For simplicity’s sake, let’s leave ordinal measurement aside.

Measurement (a) is probably easier to calculate than (b), as we can neglect differences between colors. I suppose that (b) will involve some kind of internal consistency score.

Any ideas?

PS: Yes, this really has a business application.

UPDATE

Two solutions for problem (a):

1. As suggested by my friend Chris, take the square of sums divided by the sum of squares, or in Excel-speak =(N^2)/(SUMSQ(n1:ni)); this will give you an absolute diversity value between 1 and N, which can then be normalized to the scale 0 to 1 by the formula =((N^2)/(SUMSQ(n1:ni))-1)/(N-1)
2. The solution I came up (during a visit to Art Basel yesterday) was to take the geometric mean, normalized to 0-1 and inversed. In Excel-speak: =1-(SQRT(SUMSQ(n1:ni))-SQRT(N))/(N-SQRT(N))

At the extremes, both solutions are the same (0 for no diversity, 1 for full diversity). In-between those extremes, solution 1 is more conservative than solution 2 – you need more diversity to get a high score. For the example above, solution 1 gives you a diversity of 2.2%, solution 2 one of 48.6%. From this perspective I prefer solution 2, but I might have miscalculated the scaling in solution 1.

UPDATE 2

As I thought, I used the wrong scaling approach for solution 1. It’s easier to just inverse the value (sum of squares divided by square of sums), then it’s very similar to solution 2 (with the difference that it’s open-ended, and 100% can only be reached with N -> infinity).

Overheard in the corridor: What does it say about you if you accidentally call your Blackberry “Blackmail”?

From colleagues in various companies, I hear a lot about banned internet sites – especially Facebook and YouTube. While I understand managers’ reaction to limit employees’ opportunities to pursue work-unrelated and seemingly unproductive activities, it still is shortsighted:

• Firstly, these tools can and are used for work-related activities (Facebook for networking, Youtube for marketing activities, etc);
• secondly, even work-unrelated activities can prove productive in the long run (otherwise work-life aspects wouldn’t be considered as part of a companies’ reward strategy);
• thirdly, limiting employees’ activities is an unhealthy sign of a companies distrusting their employees to do the job they are hired to do; how time is spent on the job should be regulated via the performance management system.