Sometimes we all get too caught up in perfection. If you work on that report 10 minutes longer, will it be closer to perfect? If you search one more time, will you find every single mention of your brand or company across the Web? There’s a point at which additional effort isn’t worth the reward.
I often find that there is a misconception about the purpose of measurement. Whether in PR or social media or physics, for that matter, the purpose of measurement is to reduce uncertainty. Note the word “reduce.”
Purpose of Measurement Is to Eliminate Uncertainty
Even reports and studies performed by the best and most well respected research firms include some level of error. How often have you seen data reported as a percentage with a +/- following it? This is error; it’s uncertainty. And guess what? It’s normal, and it’s OK.
There are few things in this world that we can measure exactly. There are plenty of things we can measure with 90 or 95% certainty, though. In PR and in social media, knowing your audience or your success with 95% confidence is all you need. It would take a lot more resources to reduce error further, and what will you really gain?
You could spend your entire workweek trying to find perfection, but will the insights you find be any richer? I doubt it. If you miss one more post in which your fans praise you, do you miss a lot?
Let me be clear: measuring has to be accurate. I’m not advocating half-hearted efforts that produce bad data and incorrect insights. I do believe, however, that there is a point at which measuring any further costs more (time, money, etc.) than it’s worth (value of information).
Of course, what it is that you are measuring will change the cost of incremental measurements as well as how important it is to reduce uncertainty. Doctors, for example, need to reduce uncertainty in their findings far more than you or I might when beginning a new campaign. And reading tweets from one more day may not cost you much at all.
We all need to learn when to say when.
Do you have tricks for reducing uncertainty without increasing costs? How much error is too much for you?
[reus id=”6″][recent posts]