How do businesses know what they should do? Some rely on experience. Some just guess (with elaborate justifications, of course!). Most seem to rely on a combination of the two. But modern business strategists are searching for something a little more reliable: data.
They are looking to be able to provide a solid foundation to business decisions that extends beyond looking into the past and guessing into the future. Most companies are already data-driven to some extent. But it’s not yet been able to outshine traditional approaches to strategy. In this blog, I’d like to explore why artificial intelligence is so important to delivering a scalable data and business strategy that is founded on more than experience and guesswork.
A data strategy determines what data you’re capturing, how and for what purpose. Dataversity describes it as a “set of choices and decisions that, together, chart a high-level course of action to achieve high-level goals.”It’s inextricably linked to business objectives, which form the north star to which the what, how and why of data must look. The holy grail of data strategy is to be able to use data to determine the best course of action, creating a virtuous cycle of data strategy.
What is AI and how is it different from analytics and machine learning? There’s a helpful rule of thumb from data scientist David Robinson for defining the different contributions of the various branches of data science. (He notes that this is oversimplified, but it’s a useful heuristic for navigating how we think about them).
So, you can see how different marketing channels bring in leads of different values or determine which salespeople have the highest close rate.
You might be able to use an algorithm to predict that people who visit your website at least three times have a 50% chance of buying something or to determine that if you observe a particular set of bank transactions there is an 80% chance that it is fraudulent activity.
This might be anything from recommending a route on Google maps (delivering an action that will achieve the desired result) to chatbots that can solve customer issues, Netflix recommendations and automated financial investment/portfolio management.
In the context of a data strategy, then, we could say that AI produces business systems capable of making intelligent sets of choices and decisions.
As the introductory quote implied, strategy revolves around clear rules about what a company does and does not do. Namely: what intelligent decisions should a business take? While analytics and machine learning provide insights and predictions, there needs to be a human to join the dots and decide on the next-best action. This makes human decision-making capacities a key limit on the power of your data strategy. It makes it difficult to scale.
Drawing on the quote in the introduction, AI can help to provide a clear set of choices that define what the firm is going to do and what it’s not going to do.Humans simply cannot hold enough information (or stay awake long enough!) to bring their experience and intelligence to every corner of the business. They need to prioritise and triage. With AI, this is no longer the case. Theoretically, every feedback loop, decision point and strategic junction can be brought within the purview of a scalable intelligence that can make decisions independently that maximise business value.
This doesn’t mean that there will be no human oversight, only that the emphasis shifts: rather than human oversight being a limitation on execution and scale, it takes on more of a conductor role, keeping the beat and watching for anything that is playing out of key! In practice, of course, companies will start small and see how it goes. But the promise of AI is game-changing: to scale intelligence across every corner of the business, beyond human limits.
Ultimately, AI has the potential to create ‘virtuous cycles’, in which intelligence breeds better data, which enables better products, which means more users, which means better data and so on.
This was possible before, but never at such scale and independent of human decision-making.
But AI is hard. You need a rock-solid foundation of data excellence. Monica Rogati developed the idea of the data science hierarchy of needs:
You’ll notice that AI is at the top. Trying to do AI before you’re ready will result in failure. As Monica puts it: “More often than not, companies are not ready for AI. ”There are tangible and intangible components that businesses will need to form the foundation of any AI program.
Once these aspects are in place, you’re ready to hit the AI prime time!
From experience building and executing data strategies for some of the world’s largest organisations, there are a few critical principles to follow.
Don’t try to boil the ocean. Get a proof-of-value (POV) going in a discrete area of the business where a successful model can be established and value measured. Once you’ve got something that works, then begin to scale
Make it measurable
This is really cool tech and it’s easy to get lost in the science and forget that this is about enabling better business outcomes. Define success at each stage, measure it, replay the business value, repeat
Pivot and experiment:
Don’t nail your colours to the wall before you have proved the concept. Keep your roadmap flexible so you can experiment and pivot in response to outcomes
Don’t try to custom-build your entire data platform. Use cloud-native tools for 80% of your needs to get the ball rolling. If you must then use more bespoke tools to fill in the gaps, but these should be the icing on the cake, not the cake itself!