According to the latest GCC CIO survey conducted by Gartner, Artificial Intelligence (AI) and Data Analytics are the game-changer technologies for GCC organizations.
Data Analytics and Artificial Intelligence (referred collectively as data analytics) among other digital initiatives often a topic that catches executives’ attention. In fact almost every organization I came across in recent years has budget provision for data analytics in one shape or another. Many organizations are expecting that something remarkable can happen in very short time and this is typically not the case.
I put a brief note outlining some of the top issues that in my view hinder the adoption of Data Analytics and its flavors (i.e BI, AI, ML and other). Issues appear to be overlooked and threaten success of the overall data analytics journey.
For convenience, we can logically link issues to phases of (i) Preparation (ii) Execution and (iii) Operation.
Preparation – at this stage organizations have not yet made significant investment, but has to make most of the crucial decisions
- Lack of understanding of AI benefits and use cases. Starting with broader alignment to business objectives (if one wants to improve operational efficiency, provide better customer service, increase employee satisfaction, or perhaps reduce risk). Defining specific metrics that would drive execution. Down to things like what kind of use case we want to consider.
- Assuming some sort of interviews with management took place and there a long list of potential descriptive, predictive analytics related or technology driven use cases (e.g. chatbot). Very few organizations at this time get to realize that they lack skills to estimate complexity and even scope corresponding projects. Many enterprises fall into the trap of judging complexity based on vast amount of tutorials, ready-made solutions with bunch of out of the box algorithms available that can do anything form complex image recognition, sentiment analysis to enterprise class reporting and compliance on top, with just a bit of fees on top.
- Unfortunately, with amount of hype around AI, poor data situation is often not understood. Very few organizations undergo data quality assessment at preparation stage or even perform proper identification of the data sources that may potentially be leveraged. Data is at the core of the analytics/AI program and lack of it in required quality has detrimental impact to the overall program.
Execution – bulk of money is spend at this stage into skills and technology acquisition
- Lack of robust data management framework (that covers data accountability, governance, provisioning, control, quality and metadata) leads to scope creeps, unmet expectations, poor level of satisfaction and in some cases abortion of the data initiative.
- Some other organizations are viewing AI/Analytics project as yet another IT project. Both types of projects are iterative in nature. However, the development cycles, required skill, infrastructure, testing and operationalization requirements are distinct. For example, software projects have concrete deliverables at each phase (or sprint), where data projects on other end are more exploratory and may result in change of assumption or objective underpinning phase (or sprint). The reason being is that unlike programming, AI and ML approaches are not declarative, but rather very subjective to the data and/or learning algorithms attempted.
Operation – at this stage organizations are expecting benefits to be realized
- Resistance over the results. Here we come with a number of reasons ranging from skepticism about accuracy (aka, my Excel can do it better), concerns about losing data/information monopoly, fears around job security to basic reluctance to make necessary changes to routine.
It appears that many of the organizations do commit investments into technology to soon and miss much of the ground work. Some of the early preparation steps might be as simple as:
- Raise awareness about AI/Analytics and how can this help individuals within organization
- Provide training to the team members to understand the subject from business and technical aspects
- Perform data quality assessment stored at the key systems
- Identify potential data sources that can enrich available data
- Draft enterprise data management framework
- Define plan of how identified gaps and use cases can be best addressed
It is worth considering professional service firms in helping organization throughout data journey. Unlike many vendors such firms don’t rush selling their yet another box to you.
