Eye-catching headlines make a compelling case for AI in business. But implementing AI is rarely the best first step for most companies.
It’s not the most exciting answer. Especially if, like me, you’ve always fancied having a friendly droid at your beck and call.
As I suggested in a previous post, our experience with companies big and small shows that the starting point isn’t actually tech at all. And that’s actually a good thing for most leaders who are building their data literacy, as the tech involved in AI is fiendishly complicated.
Through our work and research we’ve identified a common set of strategic steps that can get you to a place where AI systems can create benefits for your business. Anyone can follow these steps, and get value along the way, for example by developing a better focus on business decision making.
Strategic steps for leaders who want to get started with AI
These steps are the responsibility of business leaders, rather than technical data specialists, because they require thinking about strategy, reframing business challenges, and considering the ethics and values of your business. They require an appreciation of your company, your sector, and your business ecosystem. They’re also a powerful way of evaluating where you are as a company and coming up with new ideas.
Step 1: what level of data maturity are you at?
From energy companies to consulting firms, the businesses we work with tend to sit within four broad levels of maturity. Thinking about which stage you’re in is a good place to start, as it will help you identify the strategy and plans you’ll need to evolve your business and get it ready for potential AI solutions.
The key consideration for data maturity is decision making, in particular the role data plays in decisions. We look at data maturity through the lens of decision making because better, more timely decisions are generally the key driver of value for businesses, and data is a critical enabler.
Here are the 4 levels of data maturity:
Level 1: opinion based decision making
At this level the majority of the decisions made in your business are based on opinions. And normally the opinion of the most senior person or highest paid person in the room. Yes, sometimes a report might be involved, but generally that’s used as a justification rather than a driver of the decision. Because data isn’t required for decision making, data access and management is more often than not patchy and siloed.
Level 2: information based decision making
At this level, reports and information are used to make decisions, and decision makers actively look for objective insights. But those reports are challenging to generate or the data and information is hard to access, unreliable, or difficult to process.
Level 3: data-informed decision making
At level 3, standardised reporting and easy to access data drives most company decisions, which is reflected in the culture of the business – no data no decision. Leaders are good at turning business challenges into data questions. At this level you’d also expect to have access to live data, and be able to start modelling future scenarios.
Level 4: automated decision making
At this level, decision making is automated where appropriate, integrating a variety of live or near live data sources. Systems such as AI are used to both automate and find new insights from data. The business culture is supportive of distributed data-informed decision making, and gives due consideration to ethics and trust.
When you find your data maturity level, what's next? Evaluate which stage you’re at, then plan your evolution!
Getting started with AI means first identifying where you are on this 4-stage maturity scale, then getting support across the business to focus efforts on improvements. This is the realm of data strategy.
Effective data strategies are more than just a plan for governing data, they’re a plan for developing data infrastructure in ways that help the organisation improve performance.
Effective data strategies place equal emphasis on changes to people and culture as technology and product development, so that the organisation is ready to get value from new data-powered methods like AI.
Step 2: reframe business challenges as data questions
A fundamental driver of the value businesses get from data is improved decision making – as a part of services for customers or for the business itself.
This relies on a solid foundation of being able to ask the right business questions, and reframing these business questions as data questions.
The task of reframing business challenges as data questions is something that, in our experience, both business leaders and data specialists find challenging to do. This is because good data questions require both an understanding of the business domain and an appreciation for how data can generate useful insights.
In our work, we find that resolving this challenge is best done through collaborations between domain specialists and data specialists.
This collaboration should focus on developing precise questions that have the potential to drive actionable decisions.
A data question drives actionable decisions when it:
- Uses simple, direct language
- Is sharp and specific
- Asks the teams doing the analytics to predict or estimate something
- Indicates what type of data is needed
In the data literacy workshops we run for leaders, decision making is generally the area that they feel most confident about, until confronted with the need to be specific and precise.
Whatever your level of maturity there’s huge value in looking again at your business challenges and considering if you are reframing them as precise data questions. Then looking at what might be the best way to answer those questions.
If getting better and more timely answers for your business requires finding patterns in complex datasets, then AI could be the answer.
Step 3: consider how AI should reflect both your mission and your values
Just because you can use AI to power better decision making, it doesn’t mean you should.
As with the people we bring into our companies, AI systems need to adhere to company values, policies, and external regulations. Ensuring that new systems reflect how we want to be perceived, is as much the domain of leaders as that of technologists.
Leaders are the ultimate guardians of brand reputation, and therefore should avoid delegating responsibility to technical teams.
When algorithms and AI do stuff that is biased, creepy or just plain wrong, it blows up in everyone’s faces. But its leaders who end up losing their jobs.
Paying proper attention to the ethics of data and AI is therefore increasingly the preserve of the teams and committees that connect the technical and business domains. It’s up to leaders to provide guidance and support:
- Guidance about the values that AI systems should reflect.
- Support with training and resources to ensure teams have both capacity and capability to operationalise ethics in their work.
Again, this is an area that leaders should feel comfortable embracing. As a leader you don’t need to know how to create a machine learning algorithm. But you do need to know if that algorithm has the potential for creating reputational risk by providing biased or wrong feeling decisions.
Think about people, before thinking about AI
Moving up the scale of data maturity, asking better questions and considering ethics will help you make progress towards the introduction of AI in your organisation. But nothing will truly change in your company if people aren’t ready or willing to adopt new technologies or ways of working.
Most data projects fail to deliver value (80% by some estimates). But only rarely is this due to technical issues. Research shows that, in the majority of cases, data project failure is due to a lack of focus on people, culture, and processes.
This means that one of the main areas of focus for leaders should be addressing ways of working, improving skills, and adapting processes. It’s a deliberate practice and, the responsibility of leaders and managers, who in turn need to have the right combination of data literacy and skills to lead change.
Leading change doesn’t happen unilaterally. It’s about organisational intent. And that generally means ensuring that the changes necessary to move up the data maturity level are made clear in both organisational and data strategies. And, that change is given an appropriate level of resource and support.
Business has always been about survival of the fittest. AI and other data-powered services are transforming the business ecosystem, meaning that companies and the people that lead them need to adapt and evolve if they want to stay competitive.
One of the great things we offer here at Mission Drive is a bespoke Data Literacy workshop for leaders that covers all things AI and more. So if you’re keen to get ahead of the crowd or just want to do some digging in this area, we’re here to help you open the pod bay doors to a more data-friendly future.
