Digital Transformation: Part 2

Once, as Operations Manager I was training a new graduate joiner on a routine task. As I was doing so, I said: "Oh, I can have this automated". The third time I said this the newbie retorted "you'll have me automated out of a job!"  A month later this graduate came to me about another routine task they were carrying out.  To my surprise, they asked: "can you have this automated for me?"

The moral of this story is twofold: 1) people are keen to automate routine tasks once they have a moment of realisation, and 2) the IT team doesn't need to be the driving force or even involved in the initial stages.  One thing with hiring graduates is that they quickly become bored: also see the Rumala Sheikhani article linked below where she writes about similar experiences.

A more recent project involved pulling information from 6 databases, producing a chart and amalgamating into a single unified document.  Two databases needed a one-off alignment to a common key.  The two departments affected needed to make small adjustments, but as the overall project was viewed as beneficial to the company, there was no reluctance to accept the changes.  This project allowed the business relationship team to have a single reference document, laid out in a standardised format. Using expert system methodology, the report highlighted opportunities to enhance services to the customer. When I left the project after three iterations, it was still not static - there was a road map for the next year, but each quarter the road map was being reviewed.  As a routine, new items were added, and priorities shuffled. The new programme for code changes over the following three months would be chartered, with ‘ruthless’, ‘nice to do’ and ‘if convenient’ priorities.  Again, this was not an IT department driven project, but could not have been achieved without automated processing of the raw data.

Not only was the information presented as knowledge, and the expert-system methodology generating wisdom, the raw data was also stored in hidden fields in the new document.  This presented the opportunity to begin a time-series analysis of newly collated data points.  Given that the project had only three iterations, it was too early to start a numerical analysis of the benefits of producing the reports.  Clear benefits were to be expected, but I’m firmly of the belief that studying raw data opens the opportunities to see unexpected results.

An example of this that I was involved in, nearly 40 years ago, was a Rasch Analysis of school children’s multiple‑choice test answers.  The first step is obvious and traditional: identify pupils with a good understanding of the subject matter.  Rasch Analysis takes this a step further: it identifies the extent to which each question discriminates between pupil’s abilities: not all questions were equally hard.  Surprises then came to light when pupil scores were compared to question hardness: the intuitive expectation would be a progression from getting questions right to getting questions wrong as questions get more difficult.  This was true for most pupils: two exceptions became apparent:

  1. A weak pupil getting questions randomly right: this would be equivalent getting “how many days in a week”, and randomly selecting ‘7’ or a different answer.
  2. The strongest pupil in the cohort routinely marked the five easiest questions wrong. On inspecting their physical answer papers, I could see that they had marked the correct answers, then later changed them.  The pupil would not only have scored 100% in many science tests that year but had also clearly learnt the life lesson of “nobody likes a smarty pants!”  I kept their secret.

This Rasch Analysis work was carried out in the days before schools had an IT department and before ICT was an examination subject.  The science department had a ZX spectrum with 16k RAM: the application was written in Z80 machine code and took an hour to process a single cohort’s responses on each 40-question paper (excluding data-entry time).  Without this early digital processing, the project would not have been feasible.

Read More:

Rumala Sheikhani:

View comments on this article on LinkedIn: