From 1996 to 2020 we were a general representative and distributor of medical equipment in Bosnia and Herzegovina, carrying lines from KLAFS, Chinesport, Gima, Nihon Kohden, Enraf-Nonius and others, and serving more than 190 healthcare institutions. We ran the warehouse, the service department, the tender office, and the accounting that had to reconcile all three. In 2020 we exited distribution and committed fully to software. Today we build business systems for companies like the one we used to be.
That history changes how we read ERP comparisons. Most are written for generic wholesale: they rank dashboards, license tiers, and CRM features. None of that decides whether a system survives the Tuesday afternoon when a manufacturer issues a field safety notice and a hospital wants to know, today, which of its units are affected.
The comparison charts ask the wrong questions
Medical equipment distribution is not ordinary wholesale with more expensive boxes. The product carries a lot number, a serial number, often an expiry date, and a regulatory paper trail. A meaningful share of your stock sits in hospitals you do not control. Every unit you place creates a service obligation that outlives the sale by years. Your purchasing crosses borders and currencies, and your customers buy through tenders that demand a folder of documentation with every quote.
A horizontal ERP evaluation surfaces none of this. It tells you which system has the prettier pipeline view. It does not tell you which one can hold a consignment count at a hospital without a side spreadsheet. Below are the six requirements that, in our experience on both sides of the table, actually decide the project.
Six requirements that decide the project
1. Traceability as a property of every stock move
Lot, serial, and expiry data must live on the stock movement itself: receipt, internal transfer, delivery, return, scrap. If serial numbers are captured in a custom field on the sales order, or in a parallel spreadsheet, the chain breaks the first time someone is in a hurry. A recall then becomes archaeology. The test is simple: pick a lot number and ask the system where every unit of it is right now, and where every shipped unit went. If the answer takes more than a minute, the traceability is decorative.
2. Consignment stock that stays on your books
Consignment is the arrangement hospitals prefer and generic ERPs handle worst. The stock sits at the customer's site, but it is your asset until it is used or implanted. The system has to model stock that is owned by you and located somewhere else: countable per site, replenishable against minimums, invoiced on usage, and visible in your inventory valuation. If the software's answer to consignment is "ship it and issue a credit note later," you will be running your most sensitive inventory on trust and email.
3. An install base, not just a customer list
The day a unit is delivered and installed, it stops being inventory and starts being a relationship: warranty end date, preventive maintenance schedule, service history, parts consumed. Distributors who service what they sell live off this record. It should be created automatically from the delivery, serial number attached, not rebuilt by hand in a service spreadsheet that starts diverging from reality the week it is created.
4. Tender paperwork as a workflow, not a scramble
Tenders decided most of our revenue, and every tender wanted the same folder: technical datasheets, declarations of conformity, ISO certificates, delivery and warranty terms, references. The night before a deadline is a bad time to discover that the certificate on file expired in March. A system built for this vertical keeps the document set attached to the product, tracks validity dates, and assembles the quote package instead of leaving it to whoever happens to remember where the certificates live.
5. Purchasing that understands borders
Importing means multi-currency purchase orders, long manufacturer lead times, minimum order quantities, and landed costs. Freight, customs, and insurance have to land in the unit cost of the goods, or your margins are fiction. We have seen distributors price tenders off a cost column that quietly excluded 12 percent of what the goods actually cost to bring in. The ERP must compute landed cost per receipt, not per year in a controller's spreadsheet.
6. Accounting that agrees with the warehouse
Month-end chaos in this business has one root cause: the inventory system and the accounting system are two different systems, or one system used two different ways. When stock moves post their own accounting entries, valuation is a report, not a reconciliation project. When they do not, the finance team spends the first week of every month explaining differences nobody can trace. This is the least glamorous requirement on the list and the one that pays back first.
The requirement nobody lists: demo and loaner equipment
One more, from experience rather than any comparison chart. Distributors live off demonstrations, and hospitals expect a loaner while their unit is in repair. Demo and loaner units are stock that leaves your warehouse without a sale: still your asset, still serialized, still accumulating operating hours and wear, often the very unit a customer later buys at a negotiated discount. If the system cannot represent "our machine, at their site, not sold," you end up with the worst of both worlds: inventory reports that miss real assets, and demo units that quietly become permanent gifts. The same location logic that handles consignment should handle demo placements, with a return date somebody is accountable for.
What you can safely deprioritize
AI-labelled dashboards, industry "templates" that turn out to be demo data, and raw feature counts. A distributor with 30 people will use perhaps a fifth of any modern ERP's surface. The question is never how many features exist; it is whether the five workflows above run without side systems. Depth in your vertical beats breadth in every vertical.
How to evaluate: three drills instead of forty tabs
Skip the weighted scoring matrix. Take your own data and run three drills with each candidate system. First, a recall drill: one lot number, ten minutes, full location and delivery trace. Second, a consignment drill: count a hospital's stock, invoice the usage, trigger the replenishment. Third, a tender drill: assemble a compliant quote package for a real tender you answered last year. Any vendor who will not run your scenarios on your data is telling you something.
This vertical is the only one we sell to. Our medical equipment practice page describes how we implement each of these requirements on Odoo, and our pricing is published and fixed per phase. If you want the requirements list above turned into a map of your own operation, that is exactly what our two-week operations audit produces.
We ran this business for twenty-five years. The system we wished we had is the one we now build. If any of the six requirements above made you think of a spreadsheet your team maintains by hand, that spreadsheet is the project.