Three things you’re buying, not one
When somebody quotes a connector, they’re quoting the first of these. The other two arrive anyway.
You’re buying the build: the mappings, the transforms, the error handling, and the reconciliation logic that decides what happens when the two systems disagree. That’s the number in the proposal, and it’s the one you can negotiate.
You’re buying a maintenance obligation with a schedule you don’t control. Acumatica ships two feature releases a year and the commerce connector changes in most of them. A hosted platform ships whenever it likes. Every one of those is a day where somebody checks whether your mappings still hold, and occasionally a week where they don’t.
And you’re buying a dependency on a person. Someone has to understand your item master, your pricing rules, and the API well enough to tell a data problem from a code problem at speed. On the projects that go wrong, that person left and nobody replaced the knowledge.
What actually drives the number
Two connectors between the same two systems can differ by a factor of five, and the difference is almost never the API. It’s the shape of what you’re moving.
| Criterion | Cheaper when | More expensive when |
|---|---|---|
| PricingThis one line moves the number more than any other. | One price per item, or a price class the platform already models. | Contract pricing per account, volume breaks, and a price that has to be right at the moment of ordering rather than overnight. |
| The item masterNobody discovers this until they export it. Export it early. | Clean, consistent, one unit of measure, no variants. | Matrix items, kits, multiple units, and descriptions that have been edited by hand for fifteen years. |
| Inventory | One warehouse, one number, and a tolerance for it being an hour stale. | Multiple warehouses, allocation rules, and a promise date the buyer will hold you to. |
| Direction | One way. The ERP publishes, the storefront displays. | Both ways, which means you need a rule for what wins when the two disagree, and someone has to decide it rather than discover it. |
| Failure handlingThe difference between a script and a system, and it’s most of the engineering. | Retry, log, and tell somebody in the morning. | Orders that must not be lost, payments that must not double, and a queue that survives an outage. |
When custom is the right answer
We build these, so treat this as an interested party being straight with you rather than as neutral advice.
A custom connector is right when the native one has a documented limit that sits on your critical path, and you’ve confirmed it’s a limit rather than a configuration you haven’t found. Acumatica’s Shopify connector syncing a default price where you need an effective one is that kind of limit. So is a pricing model the platform has no concept of.
It’s also right when you’re running something the ERP has no connector for at all. Acumatica ships native integrations for Shopify, BigCommerce, and Amazon, and nothing else. If you’re on Shopware, or trading through a retail network, the build is the only option and the question is who owns it rather than whether to do it.
It’s the wrong answer when the real problem is data. A connector can’t fix an item master that disagrees with itself, and building one on top of bad data buys you an expensive way to publish the same inconsistencies faster. Clean first, then integrate.
The middle path most people miss
Between the native connector and a full build there’s a layer that’s cheaper than both and gets skipped because it isn’t a product you can buy.
Acumatica’s Entities form supports custom field mapping and filtering per entity and per direction, including user-defined fields and attributes. A field the standard mapping ignores can often be carried without writing a connector at all. It doesn’t cover every entity, so it’s worth checking rather than assuming, but we’ve seen it remove the reason for a build more than once.
The other middle path is scope. A connector that moves orders, stock, and customers, and leaves pricing to a nightly export, is a fraction of the cost of one that does all four in real time. If your pricing changes monthly rather than hourly, buy the cheaper one and spend the difference on the item master.
Checked against: Acumatica help, Entities (BC202000): import and export mapping and filtering per entity, Acumatica developer docs: the Commerce Framework for building a connector
Four questions that price it
Put these to whoever is quoting, before they quote. The answers change the number more than the day rate does.
Which fields are moving, in which direction, and what wins when they disagree? A connector with no conflict rule has one anyway, and it’s whichever system wrote last.
What happens when the ERP is unavailable for an hour? The answer is a queue, a hard failure, or silent data loss, and only two of those are acceptable.
Who runs the regression when Acumatica ships 2027 R1? If the answer is us, that’s a line in your budget twice a year. If it’s you, that’s a person.
What does the second storefront cost on this connector? The honest answers range from nothing to a second build, and it’s cheaper to know now.
The wider picture
This page answers one narrow question. Acro Commerce covers the strategy around it.
Common questions
- Is middleware cheaper than a custom connector?
- Usually to start, and the saving is smaller than it looks. A platform like Celigo or Patchworks gives you the retry, queue, and monitoring you’d otherwise build, which is real value. What it doesn’t give you is the mapping, and the mapping is where your pricing complexity lives. You still pay somebody to configure and maintain it, plus a subscription, so the honest comparison is middleware plus configuration against a build, not middleware against a build.
- Can we start with the native connector and replace it later?
- Yes, and it’s often the right sequence. Launch on the native connector, find out which of its limits actually bite you, and build only that. Sellers who go straight to a custom build frequently rebuild things the native connector did perfectly well, because nobody had used it long enough to know.
- What is the cheapest thing we can do to reduce the cost?
- Export your item master and look at it. Most of the variance in connector quotes comes from data condition, and most of the surprises in a build come from fields that turned out to hold three different conventions. A week of cleanup before anyone quotes is the highest-return week in the project.
Last updated 2026-08-24. Facts on this page last checked against source 2026-08-24.