Every AJM engagement starts the same way. Before a site visit, before a meeting with the operations team, before any presentation about what we do — we ask for a data export.
One file. Whatever the client can pull without writing custom code. Usually it is inventory and procurement. Sometimes freight. Sometimes the general ledger by cost centre. We work with what exists.
Step one: profile the data
Before we ask what the data says, we ask what the data is. Row counts, time range, column definitions, missing fields, available joins. This sounds administrative. It is not.
The way a company’s data is structured tells you almost everything about how the company makes decisions. If inventory and purchase orders live in separate systems with no common key, finance and procurement are not talking. If freight invoices are stored by carrier but the general ledger books them by cost centre, nobody has ever seen total freight cost by lane.
Step two: reconcile to source
Once we understand the structure, we tie the export back to a number the client already believes — usually a line on the P&L or a balance sheet figure. If the export reconciles cleanly, we trust it and move forward. If it does not reconcile, we stop and find out why before we draw any conclusions. A $68 million inventory finding is only meaningful if the $68 million ties back to the number the CFO is looking at in the financial statements.
Step three: look for the joins that do not exist
After profiling and reconciling, we look for what is missing rather than what is present. Which data sets should be joined but are not? Purchase orders against inventory receipts. Car location logs against load schedules. Freight invoices against carrier contracts.
The rail car engagement — $10 million per year in recoverable utilisation — came entirely from joining car location data to the load schedule. Two systems. Both working correctly. Never connected. The answer had been sitting in the data for years.
Step four: stress the recommendation before it leaves the building
Once we have identified a finding, we model it in simulation before we write the memo. A digital model of the client’s actual operation — real demand patterns, real lead times, real exception rates — run against the proposed change. If the change holds under stress, the memo says so and names the number. If it does not hold, the memo does not go out.
Why we do not start with a workshop
The standard consulting engagement begins with stakeholder interviews. We do not start there. We start with the data because the data does not have a political position. The operations team will tell you what they think the problem is. The finance team will tell you what they think it is. The data will tell you what actually happened. That is a different and more reliable starting point.
If you have a data set and a sense that the number in it does not match what the business should be producing, that is exactly where we start. amadden@ajmsolutions.ca
Arthur Madden is a CPA CMA MBA ICD.D and founding partner of AJM Solutions Inc., based in Calgary.
Related: The financial data your reports were never designed to show — step three of the diagnostic, where the largest recoverable losses tend to appear. Also: why a fixed-fee pilot is the right way to start.