Business intelligence was supposed to bring about intelligent decisions by giving analysts and the C-suite greatly increased transparency into the health and performance of their companies. Business KPI dashboards were supposed to present the most important information in the most digestible way possible. And actionable insights were supposed to leap right off the page.
Well, actionable insights don’t discover themselves and the most obvious “insights” gleaned from business KPI dashboards often aren’t that intelligent.
Learning the wrong lessons from KPI dashboards
Correlation of two sets of data, of course, does not equal causation, even at over 99%. Visualization gives you a picture, but not the whole picture. Context and logical plausibility are required to put both visualizations and their underlying data into proper context. Without understanding, you won’t be able to turn data into the actionable insights needed to address real-world problems in public health or business.
These are all the problems with bad analysis exemplified in the above proposal. Combine them, and you have the dashboard hell many analysts are forced to live in.
Manually monitoring more metrics on dashboards is much more likely to result in dashboard overload than a more complete view of your company’s performance. This is because data analysis skills, and access to very granular data are required to actually move those KPIs in the direction you want. There are more people using these dashboards than know how to use them.
Just because two metrics are strongly correlated, doesn’t mean they are actually related. Before jumping to conclusions, you first have to understand what the two metrics actually mean and where the source data for the KPIs actually comes from. The bogus correlations which can get all the attention end up drowning out the real correlations hiding in your data, the clues you need to react.
Stale data = missed opportunities
The only thing worse than making bad snap judgments from data is making bad judgments from old data. As AI analytics vendor Anodot explains, business KPI dashboards almost always use stale data which is useless for real-time analytics and decision making. For many companies, major wins and losses can occur within a day, and often within a few hours. If your dashboard tools rely on data contained in a weekly generated report from IT, you’re never going to be able to react fast enough.
Artificial intelligence powers business intelligence
To find all the clues, you need to be monitoring every metric which goes into your KPIs in real time. That’s a task dashboards simply can’t perform, because they rely on the user for the actual analysis and there’s only so much information a single analyst can process.
Business KPI dashboards present lots of information, but little insight. Lost in the cacophony of charts, numbers and text are indicators something’s going wrong, such as:
- A runaway bidding algorithm which places too many high bids on a large demand side platform (DSP), resulting in massive overspending.
- A sudden spike in failed ecommerce transactions due to a broken API for a single payment provider.
- A large drop in web traffic and purchases (and thus revenue) for a specific product caused by a competitor’s change in ad bidding strategy.
Each of these scenarios involves both a top-level KPI and a much more granular metric. This in fact is how real business intelligence is extracted: finding the specific which explains the general. Yet that can only happen when both the top-level KPI and all the individual metrics are monitored in real time.
Large-scale monitoring is impossible with business KPI dashboard tools because they can only present a few metrics at a time to the person using the dashboard. Even if you could hire thousands of analysts to manually monitor each individual metric, why would you subject your personnel to this?
AI analytics built on top of machine learning can produce the intelligent correlations you need to respond in real-time to sudden business events like the three described above. Instead of spending an eternity in dashboard hell manually monitoring only a few metrics, outlier detection solutions which form the core of AI analytics are able to find the abnormal signals in every single one.
So, as AI develops and becomes even more integrated into business processes, it’s likely that your business will be thankful for investing in automated intelligence led business practices to save your company from financial catastrophe by finding the crucial signals business KPI dashboards miss.