AI Series – Part 2: Data Analysis

2) Data Analysis

I previously discussed “Data as an Asset”. Data analysis and preparation are best approached as a company cultural mindset. Standardized systems, processes of decision making and a CoE to manage the data with the same focus and fervor that you would for the services and products you sell. An example of an action (which can become a rule) would be to apply the concepts of proper ‘metadata’ management, naming conventions and auto labeling (Jeff Jonas) to better aggregate your data sets.

Another example is to assess your current data models and put the data in the Right Object to feed the output for AI and other items. Each process lifecycle is fed by a data pipeline that may come from multiple systems. I cover this in the 6 Components of AI article. Its short and gets to the point.

I have seen some companies putting data, triggers, etc into a supportive object like the Account object that should be in a more actionable object (ie. Opportunity or custom object). There are general “supportive” objects and there are “actionable” objects that push a process forward. Know the difference and even change your model (painful I know) if you need to for greater flexibility and to build a more antifragile system. Now is the time to do it before your data really starts to grow.

Once you begin the process of analyzing your data (data pipeline, supportive data, actionable data) per industry standards, division, system location, etc, use your CoE team to find a measure in points (per value) for data per lifecycle. Find the gaps where data is needed to support the automation and AI initiatives. For example, ‘Data Quality Analysis Dashboards’ can provide visually driven dashboards to identify deficiencies in your record data from key standard fields.

Just a note: This data can be sourced from many locations: (1) external to the company business or multi-source, (2) Real-Time data mentioned above as actionable and (3) the company data or proprietary data that is not shared externally.

Just a quick note on system design. You may have to add additional levels of data in your Field Service UI for input as some assets are not equipped to provide the right kind of data needed to make actionable decisions. Use third party checklists and flows (ie. Youreka) or custom Dynamic Checklists can work with the asset to provide a balance of value, effort and automation bringing in some deeper data sets to apply to your models. More on this later.

Now to take action…

The next article will cover…

3) Data Collection and Preparation