Wednesday, December 15, 2004

Identifying high performers

Heather, on her "Marketing at Microsoft" Blog wrote a great piece recently on the work she was doing to define a recruitment strategy at Microsoft. Specifically it discusses how she is working to identify the top performers from the average ones so they can focus on identifying and attracting those.

Given some previous experience I decided to add a comment. I republish it here to share some of those ideas:

I've done this sort of exercise in several firms over the last 5 years - let me show several interesting hints.

1) Often the best place to start is your corporate HR database. The more information here the better but given your size, even if the data is not complete you can generally get some statistically relevant results. One measure is to track rate of progress over time (who gets promoted the quickest, constantly). CHAID analysis on big groups can produce interesting results (you will have someone in market analysis with the tools and knowledge how to do this)

2) For recent hires you may well find referrals perform the best though looking longer term might reduce this impact. This is due to the initial process (often called socialisation, where the psychological contract is cemented). Referrals often have the referrer as a 'buddy' to help them through this process increasing their initial performance

3) Many of the 'best places' for hires will also produce the most average hires (sorry, no easy answers). You don't want to ignore the poorer performers - you want to identify what were the differences between the total population and the population of hi-po.s

4) Ensure that you really understand how you will classify talent. A good team depends on a mixture of people, and performance is often due to the environment that people work in (team dynamics etc) rather than the individuals (there are lots of sports examples to show the teams full of the best players often aren't the ones who win). 360 review data often shows interesting information. Are you promoting people because they are great at managing upwards yet their reportees don't rate them well.

5) Analyse your selection data over time. Is what they are tested on really a good predictor of success? Do regression on review data with selection data.

6) Try and track the high performers opinions. The employer opinion survey can produce some great information though you will need an identifier - for example 'have you been on X training', where X is provided to high performers only. One thing you'll notice is that high performers have a stronger interest in corporate performance and leadership whereas many will have more interest in their immediate team and what their role is. My guess is that it is the attitudinal stuff that is a good indicator - do their values match Microsoft's

7) When you find something drill down. Why is that occurring? Develop stories and test them. Eg, with one firm I found that their best recruits (based on speed of progression) came from the London School of Economics (it wasn't on the firm's list of schools they presented at). The reason? The LSE has 70% or so overseas students (as well as a great academic team) and it was the intercultural experience that was the key factor. We then looked at other high-performers and found many were the sons / daughters of families who had changed countries many times in their childhoods.