It seems at any one time there are dozens of AI ideas (both within the company and from third party vendors) that advocates say will revolutionise your business.
With all this noise, it’s more important than ever to pinpoint the right areas and opportunities to invest in, to be able to articulate a clear business case, and to demonstrate quantified outcomes after the fact.
But how can you determine which initiatives to pursue and which to put on the back burner?
How can you feel confident that you’ve made the right choice?
And when the time comes to justify those choices, what numbers are you going to call upon to back you up?
When it comes to that last question, business leaders will typically ask data analysts for some KPIs to bolster their case. And rightly so - when done well, KPIs are a great way to demonstrate the impact of an initiative.
Unfortunately, the process of coming up with a good set of KPIs is often not given the attention it deserves. It’s frequently a bit rushed – something that only comes to mind right before some supporting evidence is needed – and this can lead to people grasping for easily measurable but superficial KPIs that don’t truly reflect an initiative’s impact.
It can also mean that KPIs aren’t defined precisely enough, resulting in ambiguous terminology that causes KPI review sessions to descend into arguments about what exactly ‘Revenue’ means and how it should be calculated, rather than focusing on what the numbers mean and what actions to take.
Having a structured approach to defining KPIs can resolve many of these issues, but what if this structured approach could also help you evaluate and qualify AI initiatives with confidence? In this blog we’ll look at how to leverage KPI Trees to understand how any AI proposal that crosses your desk ties back to your business’s strategic goals, giving you a solid basis to target the initiatives that are going to make the biggest difference.
The KPI Tree is a framework for developing strategic KPIs developed in 2011 by Bernie Smith, a leading KPI consultant who has worked with major financial services organisations including American Express, HSBC and UBS. It’s a great collaborative methodology for ensuring that KPIs are defendable by directly tying them to measuring strategic outcomes, as opposed to being isolated measures with nothing to back them up.
The basic premise of the KPI Tree framework is to start out with your business’s high-level Strategic Goals, then (over the course of a couple of 2-hour workshops) map out several Outcomes that will help you achieve those goals, Actions you can take to achieve the desired outcomes, and finally some Measures you can collect to track progress towards the completion of those actions. This flow can be depicted as a tree diagram with the strategic goals at the top and the measures at the bottom:
.jpg)
When thinking about which AI solutions are best for your business, one crucial thing to remember is that AI is not an end in itself, but a means to an end – a tool that enables your business to achieve its strategic goals. When correlating this with the available categories in the KPI Tree, it becomes clear that AI initiatives would fall into the Actions layer – the set of things you will do to bring about your desired outcomes.
If your organisation takes the time to set up a KPI tree – mapping out your strategic goals and the broader outcomes that will help you progress towards those goals – it’s then simply a case of validating whether the proposed AI solution would help to bring about one of those desired outcomes. If so, the solution can be added to the tree as an Action along with appropriate measures of success. If the solution does not map to any existing desired outcomes, it may not be something worth prioritising over other initiatives at this time.
.png)
This approach provides clarity on the intended value of AI solutions and makes it easier to prioritise them based on how closely they align with your business’s strategic goals. It ensures that each proposed AI project is explicitly tied to measurable business outcomes rather than being pursued simply for the sake of adopting AI.
Let’s say you’re the CIO for a major electricity supplier, and somebody has proposed developing an AI product that monitors critical energy infrastructure. The solution attempts to predict asset failure to enable preemptive repairs.
You know that one of your organisation’s major strategic goals is reducing operational costs, and your analytics team has already worked with you and other business leaders to develop a KPI Tree that highlights three key outcomes that will help achieve that goal: optimising workforce efficiency, optimising the reliability of critical infrastructure, and reducing customer service costs.
When considering the proposed asset maintenance solution, it is clear that it naturally aligns with the outcome of optimising infrastructure reliability, as being able to predict and mitigate faults before they occur will reduce emergency repair costs, extend asset lifespan, and improve overall network reliability.
By structuring the initiative in this way, it’s clear how an AI-driven predictive maintenance system contributes to the broader goal of reducing operational costs. Additionally, the Measures provide a quantifiable way to assess whether the AI initiative is delivering the expected results. This approach provides a strong, data-driven justification for investing in AI, ensuring that the initiative delivers tangible business value rather than becoming just another tech experiment.
Aligning AI initiatives with strategic business goals through KPI Trees can be a game-changer in determining which AI investments are truly worthwhile, but it does require some up-front thinking to develop the initial KPI Tree capturing your organisation’s strategic goals, desired outcomes and existing actions aimed at reaching those actions.
If you want to branch out into using KPI Trees but feel stumped on how to get started, Mesh-AI has analytics experts well-versed in effective KPI development and measurement who can help you.
.jpg)
