Author: Anne-Flore Elard, Jan 6 2021
As a continuation of my previous post Leadership in Data and Analytics, there are a few frameworks that can be useful. I will go through three of them: 1 – data value, 2 – data-driven culture assessment, 3 – capabilities
In companies I worked for, decisions were rarely based on data. They were often taken based on conviction, instinct, experience or knowledge. Sometimes data was not extracted in a way that made sense or that could effectively support the decision process. Sometimes data quality was lacking and data was not trusted.
Yet, the growth of data has been exponential and just as miners dreamt of extracting gold from those sierras, many analysts or data scientists are attempting to extract value from data. And there should be value in it.
Source: T1 – Big Data Characteristics, Value Chain and Challenges, AU – Rahman, Musfiqur. Accumulation of data over time in exabytes
But what’s the value of data? Accumulating data is a starting point, accumulating data of good quality and in an easily accessible and auditable way is better. Just as having more mines but mines that are accessible. This is often a challenge for semi or unstructured data. At this stage, value is still far away.
To become valuable, data needs to be transformed into information, and – even more valuable – into knowledge. Knowledge can then be used as a competitive advantage or as decision support (provided it is timely and correct)
The efforts to get from Data to Information are greater than the ones from Information to Knowledge. Big Data does not make it easier, it makes the required efforts just as Big. I love the way Roberto Rigobon* summarised that in the Billion Prices Project: ‘Big Data – Small Info Syndrome’. It is synonym of Big Cost – Small Value. Unfortunately, many companies have the Small Info Syndrome
So what to do? To get value out of Data and Analytics, company objectives (assuming they are defined) need to be translated into decisions that leaders will need to take, and those mapped to the pieces of information required. If a decision is recurring, for instance what do I promote when, then information needs to be transformed into knowledge by a second layer of analysis based on information. The definition of Information and Knowledge that will help the company is the guiding star.
The inflation of data without an organized and systematic value extraction is also a symptom of companies that understood that data can be a differentiator but whose data culture is immature.
There are five axes (framework #2) that can be used to understand how data-driven the culture of a company is: 1. reporting – how orchestrated are the efforts to set up analytics? Is data made available for analysis and basic reporting? (as opposed to adhoc uncoordinated requests, spreadsheets). 2. analysis – are predetermined reports provided by the analytics team vs. many separate managers? Is the data analyzed at a deeper level? 3. is there a care for optimization? Do automated actions (vs. manual) follow the delivery of information? E.g. Amazon supply chain detects that a label was pasted wrong, managers automatically receive a message and get it corrected. 4. Are analytical features integrated in products or services? 5. Are information and knowledge derived from Analytics a competitive advantage (innovation)? From 1 to 5, place the cursor on where the company is
That leads to the last framework I wanted to cover in this post. The depth of the analysis provided in a company reflects its capabilities (there can be different pockets of analysis at different levels in dysfunctional analytics organizations). There is usually a consensus on three levels of depth: descriptive, predictive and prescriptive. KPIs and metrics are descriptive, data science is predictive, real-time complex analytics is prescriptive (shapes actions and perceptions)
Finally, the data analytics process should follow the following steps: information requirements, data collection, preprocessing, scripting, evaluation, communication, automation and continuous improvement
*Roberto Rigobon, MIT Sloan, NBER, CSAC