The first programming class I ever took, I was introduced to the acronym G.I.G.O. For those of you who have never had the distinct pleasure of such a course, it stands for Garbage In, Garbage out. The premise is that if you input bad data, the only result you can get out is similarly flawed.
Traditional HR does not have a heavy emphasis on data or statistics. We ask finance for numbers, or we run a report and generally trust what is there. Or we like and trust people, so we ask them for numbers. However, as we strive to strategic if not enabled business drivers, we need to pay more attention to where our numbers come from and how those data sources affect the baselines that we use to measure progress against.
I recently had a client provide me with baseline cost data that clearly showed a disproportionate spend on one particular area of operations. Without getting into a lot of detail, it turns out that baseline cost included quite a bit of misplaced expense that significantly bloated the baseline. After removing the added expense, the project couldn’t be justified on a cost basis alone. It was still critical that the client generated a true baseline, but not having an accurate picture up front really took the wind out of everyone’s sails.
The reliability of your baseline data can be directly linked to the maturity of your data analytics culture and the tools you use. There are four primary sources of baseline data listed here from least reliable and mature to most.
Conjecture, anecdote, and correlation – probably the most utilized source of baselines, this method involves making educated guesses about the metrics you use for baseline calculations. This approach is best characterized by using very small sample sizes, and applying logic. If it takes this long to do a known quantity, then this other target quantity is x% of that time. While useful for napkin calculations, this type of data is highly subjective and has credibility problems. At some point someone will question the validity of numbers and it becomes a trust game.
Survey/poll – When you don’t know something you think you should, its human nature to reach out and ask someone that you think should know. In HR terms that means asking a group of employees about the activities they perform. This can be as simple as an email or complex as a survey. When using this type of data to form a baseline you have the benefit of a larger sample size (assuming you get enough responses), but you are essentially either asking an employee how they feel about something or what they imagine something is. There is no way to validate the data and it leads to interventions that are centered on changing the way some one feels instead of changing the way they think and consequently act.
Activity logs – If you aren’t measuring something that you really want to measure, the best way to fix that problem is to start. If you wait to measure until you’ve already changed something you missed out on a lot of good data and it really didn’t help you generate a baseline. In the simplest terms, start asking people to track what you want to measure. Think of it as a naturally occurring experiment. If you want to know the data state of the current state, start logging some data. It will take a while to generate enough of a sample size for it to be reliable data, which means it’s not a great technique to use when you’re already under the gun. However, more mature organizations anticipate and proactively measure.
Enterprise systematic – Of course the best way to generate a baseline is on reliable data that is tracked systematically across your entire sample size over a long period of time. Look for sources of this type of data in your organization first, ERP, timekeeping systems, and payroll are all rich centers of reliable baseline information, but typically require set-up and configuration to get exactly the type of data you need for good baselines.
At the end of the day, having one singular source of truth for your baselines is something to aspire to. If you’ve already got it figured out, my guess is that you haven’t read enough of this post to get to this point. The reality is that as we get started with measurements, metrics, and analytics we have to develop and plan for our baseline data sources. The better the data we get in, there better data out puts we’ll be able to produce, and the more we can help our companies move forward.