Data Driven Decision Making
How inadequate data governance and poorly integrated systems hinder your organisation’s objectives setting and performance management.
The Problem
Every year we set ourselves New Year resolutions, businesses set out their strategic objectives, setting out the yearly goals and roadmap ahead. At the end of every year businesses more than individuals review their performance against the objectives / targets / expected outcomes.
At the end of 2020, and for the last few years, I received a data summary of my exercise from Strava, a fitness app, advising me a number of things I achieved over the year and asking me to share with people on social media.
What struck me is that I had increased my exercise by 81% from 2019 yet this didn’t translate to my weight as that had increased slightly. I wanted to know more thus I contacted my gym Les Mills for them to share my exercise data, however they only had 2 data points’- visits to the gym and spin classes count (as spin is the only class that required booking for).
I then started thinking which device / app has my whole data? Surely Fitbit; I wear it all the time, it tracks me everywhere including when I am at the gym. However when I log onto the year dashboard all I can see is steps, floors, distance travelled and calories burned.
Breaking this down and the outside in approach
How useful is it really knowing how many fictitious floors I have climbed? Or how many fictitious equivalent KM I would have walked, I mean really who provided them with that requirement? They surely didn’t ask me as the user of the product and data.
What I am really interested are things like:
1. Type of exercise used and calories burned per type of activity in order to identify which type of exercise burns the most?
2. Which type of activity and associated heart rate in order to determine which hits my peak heart rate?
3. What is the average daily or monthly steps?
4. What is the average time I exercise? Do I need to increase this depending on answers to Q1 and 2? Does it differ at different times of the day?
The complexity here is that if there is not one single source of data that is consistent or can talk / integrate with the other, how am I to measure success and outcomes achieved?
Depending on my metabolism, body type, gender, auto immune, diabetes etc. conditions, the answer to question One could indicate that running and cycling or spinning is best for weight loss and therefore I can make the decision that gym membership actually is not beneficial and save myself some cash.
However based on the data available, I am unable to make that decision. Therefore the best estimated guess is what I have done in the past and it looks likely that I will have to continue doing.
How is this problem affecting your business?
Data-driven and evidence-based decision making is best at eliminating personal bias, enabling lean long-lasting cost savings. However the example above is a very real issue for many organisations, which are limited in the tools they have available to move to an evidence data-based decision making model.
The problem is greatly exacerbated within medium to large corporations due to having so many legacy systems, which are poorly integrated, and not having an enterprise-level data governance. The latter, in laymans terms is defining the data points the organisation really needs vs what is available.
The legacy system issue is the single biggest challenge for medium to big banks post GFC also and it is quoted by major organisations as a bigger threat than new Fintech entering the market.
What I have seen in my experience that the CTO / CIO suite are purchasing more tools on top of these legacy systems – e.g. the buzz words, let’s plug in some AI, Predictive Analytics, Machine learning tools etc.
In my example above, if I purchase a machine learning tool to link to my Fitbit data and run some predictive analytics, the whole thing will fail. The predictive algorithm needs clearly defined sets of meaningful data points that I need to measure as described above. The data that I need is measured and governed and quality checked – e.g. exercise activity type, heart rate, calories burnt, time per activity etc.
The latter will require some re-work by the Fitbit development team before the data is ready for us to plug into predictive analytics. Not forgetting that we also need quality and volume of data, and the monitoring of these algorithms and checks needs to be taken into account.
Is your organisation experiencing similar problems seeing the wood from the trees? Are your vendors promising the world as long you purchase the latest and greatest tool?
The transformation portfolio is further complicated by the delivery methodologies – have you tried waterfall, agile or hybrid models and it is still not getting off the ground?
Perhaps it is time to consider a different type of consultancy that offers a pragmatic approach to problem solving, offering independent reviews and recommendations within a short timeframe that does not entail countless PowerPoint presentations, and will work with your organisational culture and ways of working.