The web seems awash these days with talk of “Big Data” ushering in new ways to look at the world and make strategic decisions. But for now, most of us don’t have dedicated data scientists, or Hadoop Clusters, or the financial resources to pour into these systems.
So how can we help our organizations make better decisions and design better programs with the systems and resources we have?
I recently read a book called How to Measure Anything by Douglas W. Hubbard. It’s not a new book, but I heard it mentioned at a Meetup I recently attended, and it sounded intriguing. It’s written for people who want to use data in their decision-making processes, but who lack actual training in statistics or data science.
It’s changed the way I look at the strategic role of data. Here are two simple, yet powerful ideas I picked up:
1. If it matters, you can measure it somehow.
A central premise of the book is that anything that is important enough to be part of your strategy can and should be measured in some way. Since many of us work with mission-driven organizations—or would like to—this can be a bit hard to swallow at first. After all, how do you quantify the story of child who has learned to read, or put a value on a life-saving medical treatment?
But does collecting and sharing those stories bring in donations? Does it nab you more volunteers? How does it relate to your primary metrics – people served, dollars-toward-cause, petition signatures, etc?
Consciously or not, you’ve attached some concept of value to each of those things. The point is not to say that these less-tangible things only matter because they affect the bottom line or some other metric-du-jour.
Rather, it’s to create a point of reference so that discussions about the risks and rewards of different strategies can be compared in some objective way.
Do you really need this much precision? Photo credit: Flickr user stevegroom
2. Be OK with ‘just enough’ data.
It’s easy to tie ourselves in knots on a quest for precision and completeness. We want to know exactly what the response rate will be on a new email campaign we’re testing, or exactly how many new volunteers signed up after a recruitment event. It’s tempting to scale up measurement and prediction efforts to try to reduce error.
More data means better decisions, right?
But improved accuracy has a cost. The first few measurements or estimates you make are easy to get and tell you a lot, while each additional bit of precision tells you less and costs more to acquire—more time, money, and attention.
For some critical decisions, you may actually need to be 95% certain that you’re right. Most of time, the cost of being wrong is not that high, and a lot less precision will do just fine.
The goal of measurement is not to eliminate uncertainty (which is impossible), but to reduce uncertainty to an acceptable level so a decision can be confidently made.
See if you can find a way to collect just a bit of data without investing much effort, and then ask yourself: “Is this enough information to make a decision?” The book dives into the math required for this kind of analysis, and breaks it down in a way that’s pretty friendly to statistics neophytes like me.
I’d recommend How to Measure Anything for anyone who’s interested in making better decisions based on data. Do you have a resource you’d recommend? Share it in the comments below.
