We talked about how Analytics can be the next logical step beyond just simple measurements and metrics.
Analytics can help by highlighting potential correlations between data and giving us more opportunities in connecting the dots and reaching insights.
While analytics can move things forward in the positive direction, reaching an insight is nowhere guaranteed or assured.
Often, human decisions can be complex. Analytics is not the be-all and end-all mechanism for decision-making.
Machine learning tools can do many data wrangling, cleaning, transformation, and hyper parameter tuning tasks that are complex but not particularly creative.
It would be foolish to simply throw tons of data at the algorithms and just do what the machine learning application tells you to do.
This means leaders and managers need to understand how their organizations really work, from both quantitative and qualitative perspectives.
The qualitative perspective often provides additional context for the quantitative perspective.
Insights with the balanced quantitative and qualitative perspectives will likely be the most actionable.
For many organizations, measurements are just data. Metrics are mostly ratios or simple manipulation of the measurements.
Measurements and metric may say something about what had happened, but they do not explain why.
For example, a popular service desk (SD) metric “First Call Resolution” was down 5 percentage points this month versus last. Was it because…
Did we have a major outage so the call volume went way up?
Did the end-users suddenly got to be more sophisticated and asked harder questions?
Did the SD analysts grow more lazy or dumber?
IT is a complex business and often there could be multiple factors in play when metric moves in the certain direction.
This means we need a more sophisticated mechanism than just simple metrics to help analyze and evaluate root causes.
Analytics can help.
Descriptive analytics, even with the basic statistics-based techniques, can help point out correlations that might exist within the data. Correlations, as we all know, do not prove causation but it can still be helpful.
Predictive Analytics, along with machine learning techniques, can help create models that might point out future behaviors. The predictive models are only as good as the quality and quantity of the data you use to train the models. Still, the models are a lot more actionable than simply relying on gut-feel alone.
Moving forward, finding the necessary data and applying the analytics will be essential for managing IT.
A typical workflow will show data often lead to analytics, which in turn lead into decision-making.
While that might make sense when data are what we have on-hand, and we ask ourselves… What can we do with the data?
Perhaps it is more instructive to look at the workflow in reverse.
The more critical, first question to ask is… What decisions are we trying to make?
That question leads to… What analysis or analytics output do we need to support the decision-make?
Which then leads to… What data do we have on-hand that can support the analytics? If not, where can we go to find the data we need for the analytics?
Data science experts encourage us to ask questions first before doing the analytics.
The important question is always what is the decision this for?
ITIL has a DIKW (Data, Information, Knowledge, Wisdom) model.
I like the model with an enhancement from “The Cognitive Enterprise” by Bob Lewis and Scott Lee. I think judgment is situational but necessary. [https://www.amazon.com/Cognitive-Enteprise-Bob-Lewis-ebook/dp/B018MS4WKK]
By applying structure to collecting metrics, we gain information.
By applying analytics to the information, we gain knowledge.
By applying expertise to the knowledge with the right situational context, we get closer to sound judgment.
Many organizations mistaken metrics for analytics and for thinking metrics or analytics alone will be all they needed to reach a sound judgment in decision-making.
Metrics of analytics can certainly help, but they are only the foundational pieces of a prudent and capable decision-making process.