Get OMS selection wrong and you will pay for it for years. If you are choosing an OMS by comparing feature checklists, you have already lost.
I know that sounds harsh. But in twenty years of supply chain work, I have never seen that approach end well. The SaaS demo looks slick, the custom-build proposal reads beautifully — none of that matters. Only one thing matters: what your order profile looks like. Once the order profile is clear, the OMS decision is ninety percent made.
The verdict first: OMS selection is about order profile, not feature lists
Order profile breaks down into four lines. Walk through them one by one.
First, daily order volume. Under 500 orders a day and over 5,000 are two different species.
Second, channel count. Selling only on Amazon versus running Amazon plus your own storefront plus wholesale — the order logic differs by an order of magnitude.
Third, split and merge complexity. A few hundred SKUs shipping whole orders versus tens of thousands of SKUs where one order splits into three parcels — that is not the same OMS's job.
Fourth, return rate. A seller with under 5% returns and an apparel seller at 25% have entirely different demands on the returns workflow.
To put it bluntly, the first question in OMS selection is never "does it have enough features." It is "what do my orders look like." This cuts both ways: lightweight SaaS wins on speed and cost, but its boundaries are hard walls. Custom-built wins on fit, but the price is that you own it for life.
The pitfall I witnessed: two years of customizing SaaS, then scrapped
In 2019, a home-goods client came to me. Amazon plus own store plus wholesale, all three running together. Around 800 orders a day, wholesale over thirty percent, and single orders routinely splitting into two or three parcels — with gift cards and split-warehouse logic on top.
They picked a SaaS OMS. Go-live was fast, six weeks, and it survived the first peak season. The trouble started in year two: the wholesale splitting rules kept getting more complex, the standard SaaS logic could not handle it, so the vendor built "custom plugins" for them. One plugin, two plugins, five plugins. By the end of year two there were a dozen bolt-ons in the system, every upgrade meant retesting everything, and customer service fought the order queue daily.
The last time we met, their operations director told me: "Lao Mi, we can't change anything anymore. Adding one small rule touches three plugins. We're scrapping it and starting over."
Two years of subscription fees plus implementation plus a stack of custom plugins — the money spent was no less than building it custom in the first place. The real cost was time: for two years the whole team fought the system instead of running the business. The lesson in one sentence: once your order profile is complex enough, SaaS has a hard ceiling. Every time you push past it, you are borrowing from the future.
The opposite extreme: paying for custom-built, using it as a ledger
There is another side to this. In 2021, an Amazon-only electronics accessories seller came to me: 300 orders a day, under 200 SKUs, 3% return rate. The owner listened to some big-vendor pitch and paid for a heavyweight custom OMS.
Three years later, half the modules in the system had never been opened. Order splitting? Never used. Wholesale price tiers? Never used. Cross-warehouse transfers? One warehouse — transfer to what? The operations lead's exact words: "We just use it as an expensive ledger."
The biggest cost of custom development is not the build fee — a build fee, however painful, happens once. The real cost is this: every future business change queues up behind IT, and every small adjustment goes through a change process. An order profile that simple never needed that flexibility. Every dollar of "flexibility" was wasted.
Drawing the dividing lines: which table does your order sit at?
My OMS selection framework for clients is one table. Find your seat:
| Dimension | Lightweight SaaS | Must build custom |
|---|---|---|
| Daily volume | Under 1,000 orders | Over 3,000 orders, or growing fast |
| Channel count | 1–2 channels | 3+ mixed channels (marketplace + own store + wholesale) |
| Split/merge | Simple — by SKU or address | Multi-dimensional — by inventory, SLA, customer tier |
| Return rate | Under 5% | Over 15%, or complex refurb flows |
If you land cleanly on one side, decide and move on. If you land in between, read on.
At the end of the day, every order rule in the OMS lands on the warehouse floor — on the picking carts and the packing benches. One splitting rule in the system is one extra cart run on the floor. So when you evaluate, don't just stare at the admin UI. Walk the warehouse and watch how orders actually get broken apart.
Now the math. A lightweight SaaS OMS runs a few thousand dollars a month in subscription (check the vendor's latest quote for exact pricing); over a year, plus implementation, total spend usually lands in the tens of thousands. A light custom build costs multiples of that in year one, then just maintenance after. The three-year total-cost crossover point sits right on that order-complexity line: the simpler the profile, the better SaaS looks; the more complex, the lower the per-order cost of custom-built.
One counterintuitive point: you choose SaaS for speed, not for cheapness. You choose custom for fit, not for power. Get the motive backwards and the money is gone.
The most dangerous middle ground in OMS selection: franken-SaaS
The clients I fear most are the ones whose order profile sits in the middle — and won't admit it. They pick SaaS, then customize it endlessly. Plugins pile up, the system gets brittle, and it ends up a four-headed monster: they never got SaaS's cheapness (the subscription still runs), and never got custom's fit (the core logic is still someone else's).
I have seen too many of these. It is always the same: whatever they saved in year one, they paid back double in year two.
My one piece of advice: before you sign, ask the vendor one question — "Can my splitting logic run on your standard edition without touching code?" If yes, go SaaS. If no, build it properly. Honesty on selection day saves two years of tuition.
By Lao Mi, Supply Chain Veteran (20 yrs)
