In Search of the Elusive "Best Practice"

"There is NO such thing as 'best practice'!" he will inform them in his not-so gentle manner. "There are only good practices that are appropriate under certain circumstances!"

While I tend not to be as adamant as James, I certainly agree with his thesis. You can't assume something will work for you just because it works well for someone else.

What is a "Best Practice"?

It usually begins with someone like you or me looking for a way to solve a problem. We may have to experiment a bit and go through a few false starts, but eventually we hit upon something that works. The problem is solved, and there don't seem to be any serious side effects!

Cool!

At this point, it is just a practice that works for someone. The next step on the road to "Best" is for that lucky person to tell others about it. Of course some of us are more extroverted than others, some of us have more colleagues who are willing to talk shop, and some of us wield more influence than others.

So this step is strongly affected by who discovered the practice, and how many people listen to him or her. I have no doubt that many practices that work are never elevated to "best" mainly because they are not communicated well.

Of course, the communication doesn't do much unless other people actually do something about it. That is likely to happen if they are grappling with a problem that they believe the practice will apply to. Of course, they also have to have the will and ability to experiment with this candidate "best" practice. And ultimately, they must also meet with some success, or they will discard the practice as worthless.

This step requires that the practice be perceived as applicable for a variety of people, and that it does indeed provide value to them. Then, of course, those people must also talk about it, so the cycle can repeat and reinforce itself.

Ultimately, the practice must catch the fancy of someone whom the industry deems to be an authority. (How such a person is crowned is its own story, for another day!) It is this industry authority who christens the practice "best" and promulgates it for all to behold.

In a nutshell, to become a "Best Practice", a practice must have provided value to a variety of people who have talked about it and made it widely known. And someone of stature must have been among those people.

So What's Wrong With "Best Practices"?

The whole process we outlined above has one gaping hole in it. All of the people along the way who heard of the process, tried it and found it to be worthless (or worse) are generally left out of the communication chain. We don't learn about all of the people for whom it did not work, which prevents us from knowing the practice's limitations and boundaries.

As James tells anyone within earshot, it is all about the context. The practice worked for people whose context (organization/project/technology/...?) was conducive to making it work. But because we have limited ability to contrast them with the contexts where it failed, we have little insight into why it worked. More importantly, we have little insight into why it might fail!

With this background, we can now understand that a practice cannot rightly be called "best" unless it has been tested rigorously and found to be applicable in all circumstances. I know of no practice in IT (or even in life) that this can be said of. (Even breathing is a bad idea when you are under water!)

So What Should I Do About "Best Practices"?

1. Downgrade it to "a good practice".
2. Take stock of the problems I am experiencing. Do they approximate the ones that this practice is supposed to solve? Or do I perceive that the practice might help me with one of my problems? If not, then it is merely a curiosity; nothing I need to take action about.
3. Compare my context to that of those who promote the practice a "best". Can I identify material contextual differences that might make my result different from theirs? If so, that is a caution (though it might not prevent me from continuing with the next step).
4. Conduct an experiment. Create a limited context where the practice can be put on trial. See if I can replicate others' success. If not, examine the results to see if further experimentation might yield different results.
5. If the practice yields good results in my context, then adopt it as part of my normal arsenal of practices. If not, file what I learned under the heading "Conditions where a 'Best' Practice Wasn't Best".