Analytics by design: 4 key takeaways
On July 19th, the 2nd annual Analytics By Design conference was held in Toronto. The conference is aimed at exploring the intersection between design, technology and digital transformation with analytics and technology practitioners. If you weren’t lucky enough to attend, here are my 4 key takeaways.
AI Could Make Businesses Less Smart
That’s intentionally provocative, but it is a real concern. As analytics becomes more widely used to solve problems, more and more analysts and decision-makers will rely on models built by others, and not based on their own work. This means that this group may know less about their actual business (since so much of learning is based on doing), so when it fails, they’ll be at a loss to understand why, nevermind being able to correct it.
Organizations should be giving some thought on to how to prevent this from happening - I believe a basic grounding in data and analytics within the organization makes sense and is something many companies with onboarding programs are already adding. This approach has already proven successful, for example, with cars. You don’t need to know exactly how all the elements of your car works in detail, but most people can change a tire or a fan belt if they need to. And they know enough to know when to call a mechanic.
AI Isn’t Replacing People
Broader society may think that AI is actively replacing people and taking away their jobs, but that wasn’t the feeling at the conference. In fact, it’s creating more work and jobs than it’s taking away, primarily as people learn the underlying technologies, create those algorithms and figure out how to implement them in meaningful (and profitable) ways. Additionally, as many panelists pointed out, humans will always be called upon to make decisions and remain in control, limiting what AI truly replaces to less complicated or gnarly tasks. Generally, it’ll be a long time before AI actually replaces people like in the movies.
While I think the total number of jobs and volume of work may be increasing, the nature of that work and the kinds of jobs are changing. It’s no longer sufficient to be only technically proficient or just have experience managing a team. The individuals who will be in demand, the people who one panelist called“modern-day superheroes” will have technical literacy and can understand the tools that are powering this transformation, they have the business acumen to identify problems and understand how they can be solved with technology and they have the soft skills necessary to communicate across organizations to build momentum.
Most Large Organizations Are Not Ready
Wanting to use analytics and AI and actually being ready to use it at scale are two different things. In order to be successful, one panelist believes that 3 things need to be present - total organizational alignment with top-down advocacy for strategy, a willingness to accelerate POCs to learn quickly, and the ability to create momentum with internal communities, support capabilities, and SLAs. Personally, I think those are pretty high thresholds and could cause companies of any size to delay getting started since getting alignment is so hard. I would recommend to any organization that they identify a small use case and just get started. Every business is awash in opportunities that a few motivated individuals can focus on and solve with analytics and data to then leverage the benefits in order to get that necessary alignment and momentum.
One other area that speaks to maturity is around strategy. There was a strong belief that having a separate analytics and data strategy doesn’t really work as it creates conflict (as there are now multiple outcomes to achieve) and signals that analytics is separate and distinct from the operations of the business. The truth, as I have noted in an earlier blog, is that analytics is an enabler of business strategy. It impacts what the business focuses on, how it works and how it evaluates whether it’s succeeding, so organizations need to think about analytics and AI not something to deliver, but as a way to ensure that what gets delivered is right.
Consumers Need To Get Engaged
In almost every panel, the lack of consumer understanding about what AI actually is and how it drives their experience came up as a challenge to broader adoption. The thesis was that this limited understanding creates privacy concerns and a lack of visibility into how AI works and then creates opportunities for bias (for example, why a consumer may be turned down for a loan or charged a higher price). And I tend to agree with these, even in a room of analytics and data practitioners, there are still things that are not totally understood.
Unfortunately, I thought some of the potential solutions were even less well understood and thought out, some of which being; centralized data registries for individuals, additional government regulation, and even using AI to monitor AI decision making - and risk creating an even bigger problem. I generally believe that it’s up to consumers (and citizens in general) to educate themselves, ask businesses tough questions and if they don’t like the answers, vote with their wallets, clicks or eyeballs.
Learn more about Rangle’s take on data and analytics, here.