Original article was published by Shai Albaranes on Artificial Intelligence on Medium
This is the time of the year when usually companies are undertaking their 3 or 5-years strategic planning and Orbia (www.orbia.com) is no different. As part of that strategic process, we at the management decided to give ourselves a very aspirational target (one of many) for 2025. We refer to it internally as 20–25–30, which means by 2025 we want 30% of Orbia revenues to be generated from data and/or services.
This is a huge ask predominantly because the industrial/manufacturing DNA of the company. The way our employees, processes and infrastructures are designed is like that (I’m being super simplistic just to make a point):
- A certain material is purchased and transported to the gate of one of our 142 plants around the world
- Big and loud machines in our plants take this material and process it in to another material
- This newly created material is then sold to a customer in a one-time transaction and there is no additional communication/relationship with that customer until the next time she wants to buy from us
As Orbia’s VP of Innovation and Ventures I was given the responsibility to build our Big Data and Machine Learning capabilities (AKA internally as Advanced Analytics) which should assist Orbia to achieve its 30% target in 2025.
The purpose of this blog post is to share how we were thinking to approach this challenge but since these are the early days of this effort, I would love to get feedback from people who have done it in the past.
- Orbia’s revenues in 2019 where a little over $7 Billions
- The company consist of 5 very distinct Business Groups (BGs) which are responsible for 100% of the company’s revenues. Those 5 BGs have their own management teams and each operate in a very different industry (e.g. agriculture, communication, water infrastructure, etc.)
- As was described above all 5 BGs share similar industrial DNA and zero experience in deploying and using Advanced Analytics
We all heard many times that a large number of advanced analytics transformations fail. The recent number that I heard is 85% failure rate. The 4 main reason for this high failure percentage are:
- Lack of senior leadership sponsorship and meaningful engagement
- Insufficient business adoption and ownership (transformation lead by IT or data scientists)
- Missing real change-management and capability building programs as part of the overall transformation
- Solutions developed are not engaging for non-technical users and do not focus on usability
Given all of the above we have decided to take the following approach:
- The main focus in the first 2–3 years is not going to be on the technology itself but rather addressing the organizational challenges to have the people trust and use the technology and the outputs of the models
- Work with the BGs to generate ownership, accountability and capability building
- Build a corporate center of excellence (CoE) with the following responsibilities:
- Use case delivery support — The heart of the corporate Center of Excellence (CoE) “hard” capabilities — both the methodology for delivering end-to-end use cases as well as the specialized talent required to help the BG to deliver (e.g. Data Science and Data Engineering experts). We wish to deliver 10 or 12 of these use cases in the first year
- People capabilities (education, upskilling and hiring support) — Including Orbia Analytics Academy, a 3–4 tier program (Execs, Management/translators, Analytics practitioners, Data practitioners) combining multiple formats (frontal, digital self-serve, on the job support/training, etc.) to build people’s knowledge, confidence and engagement with Data and AI as part of running the business
- Partnerships — Help BGs source, asses and manage Analytics partnerships with external providers
- Tech and tooling — work with CIO of Orbia to define the required technologies and infrastructure that need to be build
- Start with the problem/opportunity and then ask which data and models can address that. Here is a great post about it by Cassie Kozyrkov, Google’s Head of Decision Intelligence
- The first step will be to run workshops with each of Orbia’s BGs to:
- Educate on AI-ML, what it is and what problems can these technologies solve that other solutions cant. Including showing successful use-cases from other companies
- Ideate with the BG teams and generate 50–60 possible areas where Advanced Analytics can generate impact
- Prioritize those ideas based on impact-feasibility and choose 2–4 ideas that will be further scoped over 3–4 weeks
- The next step after the ideation workshops would be to pilot those selected ideas and hopefully start showing positive impact within 3–6 months. These early successes will generate positive momentum and appetite from the BGs to continue running the program
- Since we lack advanced analytics capabilities we will, In the beginning, outsource most of responsibilities and over time build the internal capabilities
Status update — as of October 2020 we already ran the first series of workshops with Dura-Line (one of our BGs) and are now in the process of putting the teams together to start executing the pilots.
Do you have experience in delivering such Advanced Analytics programs? Can you share what worked well or not so well for you.
I’m recruiting for the senior person who will lead this transformation program at Orbia. If you have prior experience in building and executing such programs please apply through email@example.com