Defining —
Analytic Feedback
The Next Evolution In Analytics Geared For The Cloud, The IoT, And The Infrastructure Challenged
Recent advances in AI have come from a changed perspective. Rather than try to teach every robot how to identify the world around it — engineers are teaching the world around it to tell them. This is being used in factories where packages use RFID chips to broadcast their size & shape. It is now used for navigation where tricky landscapes now broadcast their own helpful insights. It lowers complexity and it uses analytic feedback. Other uses will follow.
I am not a big fan of conflating AI into general analytics. It tends to amp the hype, explode the budget, and inevitably disappoint. I AM a big fan of using AI innovation to remind folks of advanced concepts and structures available to all areas of the analytic landscape.
I have written at length about feedback on numerous occasions. I don’t want to revisit the past, but lets pull forward a basic definition. It is a definition that helps distinguish feedback from data.
Feedback is data transmitted with an implied source and a receiver.
It is a very basic definition but an important one. Now we are going to add a very important adjective. Analytic. Let’s dig in there.
The etymology of analytic is interesting. It is the word used by Aristotle when he created the concept in his original writings. The word’s original meaning is tied to the term “unloosen”. Let me offer the synonym “structured”. This is quite basic, but critical.
Most feedback is unstructured. Most unstructured feedback is uncollected or unused. Meaning, most collected/used feedback will appear structured, but that is a sampling anomaly.
Collecting unstructured feedback has become more common in the last decade. But, until a meaningful structure is added, that feedback will go unused and provide little value. Most unstructured feedback sits in Hadoop repositories collecting dust or creatively playing the role of “elephant in the room”.
But this article isn’t titled Structured Feedback, it is titled Analytic Feedback. The connection of Analytic to analysis is more than just buzz-y and convenient. Plenty of collected and structured feedback is actually overly structured — call it restricted or forced. In stark contrast, think of Analytic Feedback as feedback that has been intelligently structured. Basically-
Analytic Feedback is data transmitted from a defined source utilizing an intelligent structure to make it more readily usable and meaningful to a receiver.
The definition is easy. The implementation, perhaps less so. That is actually too tongue-in-cheek. Implementation is often soul crushing. This isn’t easy. Anyone telling you otherwise is clearly a salesman.
Historically, this has been done in fairly old-school fashion. Feedback is collected, stored, and added to a parking lot for potential analysis, structuring, and usage (emphasis on potential). Assuming that an analytic team is able to create meaningful structure, that is typically added after the fact by ad-hoc or batched processes in the ADB (analytic warehouse). Sometimes those processes are added to a parking lot for IT or Dev to implement the solution “up stream” (emphasis on parking lot).
So let’s come back and kick Ethe can…
In a world of cloud, IoT, and AI innovation — we can turn this model on its head. Coming back to the opening analogy of robots and cans — what if we pushed the analytics even further “up stream”.
How would this work? How is this not just the Developer ask all over again? Am I really suggesting pushing the development on analytics into the apps, websites, and points of sale? No. Not at all.
In our earlier example, the can isn’t providing any intelligence. It is the RFID chip. The can has structure, but the RFID translates that structure for our robot. It creates the intelligence. We can leverage that model. We can create an RFID chip of our own.
Our RFID chip would be:
- attached to the source
- small/innocuous
- a broadcaster of intelligence
- cheap to implement
- cheap to maintain
It would change the old school dynamic but without breaking the infrastructure. RFID chips allow the robot’s inner workings to change, but not radically. The robot actually becomes more efficient. Analytic Feedback intelligence developer (AFID) should do the same for any data environment. Easy? No. But no longer soul crushing!
In future articles, we can develop this further. For now, lets leave it here. Analytic Feedback by definition is a higher and more valuable form of data. It would allow us to make better and more insightful use of a world of currently unstructured data. The challenge is creating the intelligent structure. While we have plenty to work through, we can use RFID as a model. Let’s see where it goes from here?
Want to get started —
