I am a big believer that data should be liberated to help the right people make the right decisions, not that the data should make decisions itself. This in some ways runs counter to the “Automate Everything” mantra that is common now but hear me out and see if you agree by the end.
Most of the people who use the analysis we do for business both in Australia and internationally are usually knowledge workers, that is people who analyse data to help make better decisions, but whether they be accountants, lawyers, engineers, manufacturers or retailers, the tools generally end up in the hands of people who interpret their business environment daily.
What each of these people that use our tools have in common is years, if not decades, of experience in their industry or a related industry, this experience has grown through experimentation, trial and error, and they have learned as we all do through failure and learning.
The problem we are starting to see emerge is automated decision making by automating the data analysis and flow to act. There are no issues where the decision is trivial or the risk is small, but I will tell you one thing for sure as I have seen it countless times, automation doesn’t stop bad things happening, it just makes them happen quicker (anyone remember the robodebt scandal?!). Now before you stop reading, I am not against automation: I run a business that builds automations, what I am wary of is automation that makes decisions!
My suggestion is quite simple and only alters the information flow slightly, allow your data to make recommendations, not decisions.
Any decision we make is bound by innumerable variables. In data analysis, what we do is take underlying data, apply context (a given number of variables) and convey the story the data is trying to tell us to the end user. The trick here is that we can only possibly allow for dozens, or even hundreds of variables, not the millions that may take place.
For example, due to some staffing issues we may need to change some shifts. Now, we could do this automatically by simply looking at availability and rescheduling people. The data is taking into consideration many different variables, but it didn’t know everything like that Mary and Geoff don’t work well together and that there is a planned transport strike on Thursday so those who need to take public transport will be affected, and so on.
It is a simple example, but one that hopefully highlights that if we simply let the data compute it and then deliver it to a human decision maker for analysis and interpretation, we then allow for all the contextually variables to enable a better decision.
Over time, the decision-making capabilities of technology will certainly improve, and this article will be destined for the dustbin, but for now, you have some wonderfully intelligent and experienced people, arm them with the data they need to build relationships with your team, rather than trying to automate decisions without them.
If you missed my earlier articles and have been convinced of the virtues of data driving recommendations and not decisions you can back
here to start your data revolution and building your data strategy, and then work towards how that data can be used in scenario planning.
Our team at PT 2.0 is always happy to talk about the virtues of using data in your business, reach out for a no-obligation chat now.
Comments