The maritime procurement problem
We conducted structured research interviews with 20 maritime fleet operators, representing a combined fleet of more than 3,150 vessels. The conversations covered procurement directors, purchasing officers, technical superintendents, and financial controllers. We were not pitching. We were listening.
What they described was consistent across operator sizes, geographies, and systems. Four structural problems, repeated almost word for word, across every conversation.
Here is what the research found.
Finding 1: The waste is quantifiable. The tools to prevent it do not exist
Procurement managers know, with uncomfortable precision, how much their fleets waste. One interviewee, managing a fleet with an annual spare parts spend of $235 million, estimated that naming inconsistency alone, the same physical component catalogued under 12 to 20 different names across vessels, drove more than $23 million per year in duplicate purchasing.
They could calculate the number. They could not stop it, because no tool in their stack gave him a fleet-wide view of what was already owned.
This was not an isolated case. Across the research, duplicate purchasing caused by part naming inconsistency was the most consistently cited structural problem, at operators of every scale.
The issue is not carelessness. It is architecture. Every vessel is managed as a silo. What sits in the hold of one ship is invisible to the team ordering parts for the next one.
Finding 2: Three-way matching fails most of the time. Every failure costs money
The three-way match, reconciling a purchase order against a delivery note against an invoice, is the financial backbone of procurement. In maritime, it fails 60 to 70 per cent of the time.
Each failure requires manual investigation: locating the original PO, tracking down the scanned delivery note (often handwritten, often from a ship agent in a port without digital infrastructure), converting currencies, reconciling partial shipments.
The cost per failed match has been calculated at $85 in staff time. At the volumes managed by large fleet operators, processing 25,000 purchase orders per month, this is millions of dollars per year in admin cost that produces no commercial value.
What makes this particularly difficult is that the data is genuinely messy. Delivery notes arrive as handwritten PDFs. Invoices span multiple currencies at rates that shift between order and receipt. A single consignment may be split across three ports. The matching problem in maritime is not the same as matching in a standard B2B supply chain. It is substantially harder.
Finding 3: Budget decisions are made in the dark
One technical superintendent described approving a $35,000 overhaul kit without knowing whether they had already exceeded his quarterly OPEX budget. Their words: “I’m often approving a requisition without knowing if I’ve already blown my budget for the quarter.” A colleague described it more bluntly: “Sometimes it’s like playing roulette.”
This is not a budgeting failure. It is a data flow failure. Budget information exists in AMOS. Committed spend exists in the accounts payable ledger. Approved-but-not-yet-invoiced orders live in email threads and spreadsheets. At the point of decision, the information that would change the decision is not in the room.
One data point captures the scale of it. Technical Superintendents reported spending 40% of their working week on data administration: downloading reports, merging spreadsheets, cross-referencing budget records. Two days out of five. Not on engineering decisions or fleet management. On manual reconciliation that produces no commercial value.
The downstream cost is familiar to financial controllers. Budget variance running at 3 to 5 per cent on a $235 million spend base produces $7 to $12 million of uncertainty that has to be explained to the CFO every quarter, usually after a 30-hour manual assembly process.
Finding 4: The data quality problem starts on the vessel, not in the shore office
Purchasing officers and procurement managers carry the weight of the matching and reconciliation burden. But the root cause of much of it is upstream: the requisition.
When a chief engineer raises a requisition from memory, hours after completing a job, using part number conventions from a previous vessel, the error propagates through every step that follows. The three-way match fails. The part cannot be found on a sister vessel because it is listed under a different name. The invoice cannot be reconciled. The volume discount opportunity disappears.
This cascade was a consistent finding. Improving procurement intelligence without addressing the data quality problem at origin produces marginal gains. In six separate interviews, fleet engineers asked directly for mobile or barcode-based requisition capture. “It’s ridiculous that in 2026 we don’t have a phone that you can shoot on a barcode.” That is a direct quote. It was not an outlier.
The numbers behind the findings
The research was not a small sample. Twenty operators. More than 3,150 vessels. The patterns below were consistent enough to quantify.
None of these are edge cases. They are the baseline, across operators of different sizes, on different systems, in different markets. The procurement problem in maritime is structural, consistent, and, as the research also showed, solvable.
Where the research led us
These four findings pointed to the same conclusion. The problems are structural, not behavioural. They have not been solved by existing PMS platforms or digitisation projects, because those tools were not built to address them. They were built to manage transactions, not to provide intelligence across a fleet.
We took the findings back to the drawing board. The result is AMOS Procure Smart: a platform designed specifically around what the research uncovered, validated on real fleet data, and built to work on top of existing systems without requiring migration or replacement.
To see how the research shaped the platform, visit the AMOS Procure Smart product page. If you would prefer to see it working on real data, book a 30-minute demo and bring your hardest matching scenario.
