Analytics is the most sought after competency in the business world right now. LinkedIn identified it as one of the most wanted skills across the globe. I have had discussions with various L&D heads who say the mandate this year is to build analytics talent within their organization. Over the course of many such discussions, I realized that while the end goal is clear for such organizations, the path is often not.
Analytics is very simply defined as the ‘language of data’. When we teach people analytics, we are essentially teaching them how to communicate with data – how to talk to data and, more importantly, how to listen to it. For those who learn to listen to data, get to hear all the secrets and insights that the data hides within.
Analytics is not an easy field to master. It requires a peculiar combination of skills – mathematics, statistics, programming, storytelling, project management and domain knowledge. This combination of skills is not easily available in the existing talent pool. Most people will have one or more of these skills but not all.
The path to building analytics competency is long. It takes time, effort and patience to get there. Shortcuts will lead to sub-optimal results. Many L&D teams realize this only after they have made some expensive mistakes.
If you are someone who is about to embark on this journey to build analytics competency within your organization or team, here are some common pitfalls that you should know and avoid.
When you are tasked with building analytics capability across multiple teams, a one size fits all approach is a recipe for disaster. Different people are at different starting points. For example, folks from the IT team may have zero knowledge of statistics but may be well versed with programming. Similarly, business teams may have good domain knowledge and decent statistics knowledge but may be unfamiliar with programming.
Since analytics requires a combination of all these skills, finally everyone needs to master all of these. However, putting all these people from diverse backgrounds into one standard training will not yield good results. Instead, a lot of thought needs to go into getting people with similar training needs into the same group and ensuring people with diverse needs don’t end up attending the same training.
Similarly, not everyone has the same end goal. For example, while everyone needs to learn analytics, not everyone needs to master Machine learning or other complex techniques.
It is a good idea to have 3 levels of training – basic, intermediate and advanced. Participation in these training should follow a pyramid structure, Everyone goes through the basic training. A subset of this goes for the intermediate training. And finally, only a chosen few, based on role, aptitude, and interest, should be eligible for the advanced training.
I can not tell you how many times we have come across training curriculums that have ludicrous expectations from the trainers as well as the participants. Here is my hypothesis for how this happens – business goes to L&D and says we want to teach our people analytics. Can you set up a training? L&D goes online to find out what people need to learn for analytics, comes up with a TOC (table of contents) based on a book or an online course, and then reaches out to training companies.
Most training companies take the TOC as-is and deliver the training. It is only after the training happens that the L&D realizes most people have not learned anything from it. Most of it has gone over the participants’ heads.
As I mentioned earlier, analytics is a complex field to master. It requires a peculiar combination of skills. It just cannot be taught in one week. It is unrealistic to expect people to master statistics, complex analytics algorithms, and programming in such a short period of time.
Analytics programs need to be designed keeping in mind the complexities of the skills that need to be covered in the training. Adequate time needs to be given to master each skill. Training needs to be spaced out over a period of time, giving enough time for the participants to absorb one skill before they move to the next.
This point is an extension of the point I made above. Analytics cannot be mastered in a short period of time. You cannot put 25 novices in a training room for 5 days and expect 25 data scientist to emerge. The ideal way to teach analytics is to give the participants knowledge on one topic, then give them time to assimilate this knowledge (preferably through case studies and hands-on learning) before they move to the next topic.
This means the training program will involve learning not only in the training room but outside of it as well. And a 40-hour training program will not involve a trainer talking for 5 days and participants listening to the trainer and trying to stay awake at the same time. An ideal training program will involve some teaching in the training room, some hands-on work in the training room and tons of hands-on work outside the training room, interspersed between regular work on non-training days.
I find it strange that often times, even business heads – people who know analytics themselves – have unrealistic expectations about the end result of the training. Training by itself will not make someone an expert. Especially, in a complex field like analytics. A good training will build the knowledge base, and show the path for continuous learning and growth. However, learners will continue to need hand holding and support even after the training period.
A smart L&D person will ensure this support and hand holding in the post-training period is part of the deliverables for the training company. This will go a long way in making the training effective and guaranteeing that the participants reach the level of analytics maturity that the organization desires.
Now that we have reached the end of the article, I want to reiterate some of the important points to consider when looking at building analytics competency in your team or organization.