Optimise Your Choice of AI Projects To Maximize Profitability

Optimise Your Choice of AI Projects To Maximize Profitability

How do you decide how to prioritise where you spend your AI investment, given so many projects struggle to generate a quantifiable ROI? 

One way to do this is to populate an ROI model with all the base data needed to develop an ROI, building future years based on your existing business plan.  Hence, you include things you have already decided will be in your future plan, such as adding additional stores, expanding e-commerce into another country, etc.  This becomes the “as is” scenario.  By including the future years, you have set the most suitable baseline for this analysis.

Then in your ROI model change only one KPI at a time in steps taking care not to get too optimistic and see how much it individually changes your pre-tax profit and free cash flow over time.  Do this for all the KPIs in the model most relevant to your objectives.  

Use the answers to rank KPIs from biggest to smallest impact.  For example, increasing full price sell through might make a better contribution to future EBIT than reducing clearance markdowns.  If you currently do three inventory turns a year in a segment where individual companies range between 2 and 5 turns, your baseline may include 3 turns, and you can simulate 4 and 5 turns and assess the impact.  

Forget the KPIs where the gain is not big enough to change decisions and rank all the remainder best to worst.  Then look for the areas to focus on to achieve the biggest EBIT increase.  Study which processes will impact the most KPI value.  Give priority to processes which impact multiple KPIs and then determine what AI improvements could be made to those processes.  

Looking at the relevant processes, if they involve inventory turn, for example, does the AI potential improvement opportunity suggest that you could achieve 4 turns but you might be pushing your luck at 5? Assess to what extent they will achieve the performance improvement you want and brief your AI vendors or your IT Department and set their expectations.

Keep in mind that critical success factors will include enlisting enough real industry experience to be sure it will work, and budgeting enough change management capacity to see the project through properly.

One key tip from our experience, if you have multiple AI projects you can implement in phases, simulate the phasing in the ROI model as this can make a huge difference to how much “new” money you need to fund these projects, compared to funding from earlier phases profit streams.  

If you don’t have access to a suitable ROI model, please contact me at brian.hume@martec-international.com.


Posted by Brian Hume
19th November 2026

Back to The Martec Minute