A week of structured diagnostic work uncovered €560K in recoverable value for an Irish distribution business — before a single AI tool was purchased.
The business was a mid-sized Irish distributor — food and ambient products, around 60 staff, operating across the island of Ireland. The managing director had been reading about AI and believed there was value to be captured. He'd had conversations with two technology vendors. Both had presented impressive solutions. Neither had asked him what problem he was trying to solve.
He engaged Acuity AI for a diagnostic before committing to any technology spend.
What the diagnostic involved
The seven-day diagnostic is structured, not exploratory. It follows a fixed process: time-on-task analysis across key roles, meeting audit, decision flow mapping, margin leak identification, and workflow bottleneck inventory. The goal is not to find AI use cases. The goal is to find where value is being lost, and then determine whether AI is the right recovery mechanism.
In this case, we spent the first two days in observation and interview — understanding the rhythms of the business, the points of friction staff encountered daily, the gaps between how processes were supposed to work and how they actually worked.
Day three and four were analysis. Day five was the preliminary findings. Days six and seven were validation with the leadership team and costing.
What was found
Three distinct areas of recoverable value emerged.
The first was workflow drift in the sales and accounts team. The business had grown significantly over five years, but its internal processes hadn't been updated to match the scale. A quote-to-invoice process that had worked at 30 staff was creating significant rework and error correction at 60. Staff were compensating through informal workarounds that weren't documented, weren't consistent, and were consuming time that wasn't visible in any reporting. The time cost was quantifiable: approximately €180K per year in salary time spent on tasks that a properly configured workflow — with or without AI — would eliminate.
The second was meeting overload. The senior leadership team was spending an average of 22 hours per week in internal meetings. When we mapped the decision-making that was nominally happening in those meetings, we found that the majority of decisions were either already made before the meeting, or were being deferred to a subsequent meeting. The meetings were status reporting dressed as decision forums. The recoverable value — time that could be redirected to revenue-generating or strategic activity — was approximately €210K annually based on the loaded cost of the attendees' time.
The third was margin leakage in supplier negotiation. The business had data on purchasing volumes, supplier performance, and margin by product line — but nobody was systematically analysing it. Decisions that should have been data-driven were being made on the basis of existing relationships and rough estimates. A conservative estimate of the margin improvement available through better use of existing data was €170K annually.
Total: €560K in recoverable value identified before any technology was purchased.
What changed, and what was recommended
For the workflow drift problem, the immediate recommendation was process redesign, not AI. The workflow needed to be cleaned up before it could be automated. Acuity AI mapped the target process, which the operations manager implemented over six weeks using existing tools. AI-assisted document processing was recommended as a second phase once the process was stable.
For the meeting problem, the recommendation was structural: a governance change to meeting formats, agendas, and decision documentation. AI meeting summary tools were recommended to support the new structure — not to fix the underlying problem, which was structural, not technological.
For the supplier negotiation problem, the recommendation was to consolidate and clean the existing purchasing data first, then deploy an AI-assisted analysis layer. The data was there; it just hadn't been organised in a way that made analysis tractable.
What this demonstrates
The diagnostic-first approach is not about delaying AI adoption. It is about ensuring that AI is applied to real problems rather than hypothetical ones.
In this case, the majority of recoverable value required process change, not AI. Where AI was recommended, it was for specific, well-defined tasks with a clear ROI case. The managing director spent seven days on the diagnostic. The alternative was spending six to twelve months on technology implementations that would have addressed the symptoms while the underlying problems continued.
If you are considering AI investment for your SME and haven't done the diagnostic work yet, that is the right starting point. See our AI strategy for SMEs page for more on how this works in practice.