THE HIGHEST STANDARD IN AI — AI GUILD #DATACAREER
Software consumes the world successfully if it captures data.
The Golden Age: Data Career in the 2020s (No 3).
That software is eating the world is a bonmot, but it is not yet understood well enough that software only consumes the world successfully if it generates and captures data.
In Europe, businesses and governments are falling behind. Internet companies may also not be fully data literate. I have seen the evidence and heard the stories.
I have encountered a mindset whereby founders, executives, and even investors focus on Digitization and miss the opportunities of the data economy. Digitization is a wrong choice.
It is not like executives don’t see the need to invest. They articulate the dynamics of the data economy and the risk of staying behind. Yet, I have not seen many companies successfully making data a core function of the product and business process.
Data practitioners suffer the most. They work on faulty data infrastructures, lack resources, and must handle data-illiterate organizations.
Let me proceed in two steps.
- Provide you with a list of things to watch out for as companies must make data generation and capture their critical focus.
- Argue why ‘digitization’ is a bad choice and how this mindset holds companies back.
Next week, in a third step, I outline three essential steps any company must take to be data literate.
How companies succeed or fail in the data economy
Since the mid-2000s, a prevalent play in Europe has been to digitize some market whenever it is ready, e.g., fashion, banking, delivery, and urban mobility. This strategy has generated new wealth and jobs, and a few companies are listed on a European or national stock market index. Yet, the playbook has not created noteworthy global players. For example, no new success has been comparable to the business software company SAP by valuation.
Companies need to catch up with the data economy. Here are eight indicators for understanding how far a company has progressed. For any company, you can research information, corroborate it with practitioners, and address it in a hiring process.
Chief Data Officer, or equivalent
The company has no Chief Data Officer. If it does, the CDO is a second-tier decider, reporting to someone else. I have encountered variations like the CFO and the Head of IT sharing responsibility. Please check if the data people have or do not have a seat at the table and the commensurate priority and resources.
Director tenure
The company hires at the Director or VP level, but the new appointee leaves in less than 12 months. The reason might be unique for each event, but the pattern is not. I keep hearing this story, and it makes me very uncomfortable. Find the current Director or Directors, note the starting month, and find out who was responsible earlier. What is the pattern?
Employee churn
Employee churn is considerable, with practitioners leaving before they can see and claim the impact of their work. In my experience, the effect becomes visible after two or more years. Employees care about promotions and meaningful work. Less than two years’ tenure is troubling. Also, this pattern has an ugly consequence: Practitioners with five, six, and more years of experience but multiple stops may not be recognized or visible as Senior. Please check the average tenure at the company.
Team size and structure
The team is too small and, typically, lacks experienced professionals. Sometimes, a bunch of juniors experiments without a team lead and principal. Three people are not a team, and a company needs several groups. Even for a startup, I would be skeptical if you could not show me at least 20+ data practitioners. Overall, the field has matured: Who is the team lead, and how many principals (key experts) does the company have?
Data as a project versus Data as a product
Management treats data not as a product but as a project, possibly even as work that can be outsourced, like IT projects. Project work means that 50–70% of the resources go to vendors, consultants, and freelancers. I don’t want to be misunderstood: They do provide value. Yet do check how much and what the company outsources. A variation of “data is a project” attitude is declaring all data infrastructure a mere commodity. By contrast, creating data value means productionizing use cases, which requires a clear rationale when partnering for infrastructure, model building, and deployment.
Data Infrastructure
There is no unified or integrated data infrastructure for the company. Maybe the company is battling legacy tech and data silos, or it doesn’t yet have an implementation plan. Companies may hesitate when selecting vendors and partners (e.g., cloud services). You want to check the status quo and the 12-month roadmap. Getting use cases to production without a unified infrastructure will be (too) hard.
Bottom-line impact
The business strategy for making data a core function for the customer and business operations is not (yet) visible. If the business cannot envision and measure the bottom-line impact, it will likely focus on research, projects, and proofs-of-concept. In other words, you won’t get that green light to move to production. While R&D adds value to a company, it does so typically if there is a product or production line. Please check because your impact over time depends on data having an increased bottom-line effect.
Innovation capacity
Only when you have the use case in production can you innovate. I don’t mean monitoring model drift but updating or changing the product or production line, a vital capacity for market success. Mature data companies have this capacity; if they do not, their efforts will be defined by catching up to the market leaders.
Digitization as a bad concept
Startups and corporates look for digitization opportunities. However, Digitization or digital transformation is a lousy concept because of its consequence: It leads businesses and governments to focus on software development and usage while missing out on the data economy. Moreover, it gives rise to the idea that you can transform analog practice or process into a digital equivalent.
“Digitization is the process of changing from analog to digital form, also known as digital enablement. Said another way, Digitization takes an analog process and changes it to a digital form without any different-in-kind changes to the process itself.”
from the Gartner Glossary
Consider the example of the PDF. Every week I have some business process that requires me to download a PDF, fill it in, print, sign the paper, scan it, and then upload it. This practice strikes me as the equivalent of passing a law stating that horses must draw an automobile.
Startup culture in Europe has been affected by this mindset. Founders and investors alike scour the market for “digitization” opportunities. Bringing something online was an early business case, e.g., fashion or collectibles. Organizing everything in an app was also successful for everything from banking to deliveries.
I consider this startup playbook legitimate, but if you have been around since, say, before 2017, then you should have invested in re-orchestrating your company through data. Some companies have done this quite successfully, e.g., Spotify.
If your company was founded after 2017 or five years after Data Scientist was declared the sexiest job, then the playbook should have identified data as critical.
You should talk to the data practitioners if you are a founder, executive, journalist, or politician. It is your best chance to understand what is happening in the engine room of the data economy — or not.
The AI Guild’s 2000+ Specialists
The AI Guild is Europe’s leading practitioner community in Data Analytics, Data Engineering, Data Science, Machine Learning, Deep Learning, NLP, Computer Vision, and MLOps.
Do you want to progress to Senior, Lead, and Director?
It takes 60 days to build your competency profile. You can find out more by booking the first conversation to gain more insights at https://www.datacareer.eu.
