ERP integration cost: what actually drives the number
Nobody publishes a price for this, and the figures that circulate online mostly do not survive being checked. So this page does the next most useful thing: it sets out the drivers that actually move the cost, the shape the money takes over time, and the questions that turn a vague quote into a real one. No email gate, no form — if it is useful, use it.
Why there is no straight answer online
Search for what ERP integration costs and you will find confident numbers. We went through a number of the most widely repeated ones. Most did not hold up, and the failure pattern is consistent enough to be worth publishing.
The source sells the remedy
The most-quoted cost-of-failure figures trace back to companies selling the fix for the problem the number describes. That does not make them false, but it does mean nobody independent has checked them.
The citation chain ends nowhere
Follow the links back far enough and several land on a gated report, a page that has been withdrawn, or a domain that no longer resolves at all. The figure survives; its evidence does not.
The attribution drifts
Numbers get re-attributed as they travel. A widely repeated "Gartner" ERP failure rate turns out not to appear in any Gartner publication we could find, while a separate figure credited to Gartner traces to a different firm entirely.
Three that do hold up, and are worth knowing: Gartner predicts that by 2027 more than 70% of recently implemented ERP initiatives will fail to fully meet their original business case goals, with as many as 25% failing catastrophically. Panorama Consulting's 2026 ERP Report finds more than a quarter of projects ran over budget and almost a quarter over schedule, with unexpected additional technology requirements the most common cause. And the long-standing 20/80 rule — that acquisition is roughly a fifth of total cost of ownership — traces to David, Schuff and St. Louis in Communications of the ACM, 2002. It is about end-user workstations rather than integrations, so treat it as directional — but the direction has held up for two decades.
The six things that actually move the number
Integration quotes vary enormously for the same-sounding job. These are the variables behind that spread. Work through them before you ask for a price and the answers you get become comparable.
1. How many systems connect
Cost does not scale with the number of systems, it scales with the number of connections between them. Wire everything directly and twenty systems implies a hundred and ninety possible links; route them through one hub and it is twenty. This is the single biggest structural lever.
2. How many entities move
Products, stock, prices, customers, orders, invoices, credits, returns. Each is a separate piece of work with its own rules, and each direction of travel is separate again. "Connect our ERP" can mean two entities or ten.
3. How messy the data is
The most common cause of overrun we see, and the hardest to assess before starting. Duplicate customer records, products that exist twice under different codes, prices that live in a spreadsheet — all of it surfaces during integration and has to be resolved by someone.
4. How unusual the commercial rules are
Price lists per customer group, minimum orders, back-order behaviour, restricted ranges. If these are configuration, they cost setup time. If they need custom code, they cost that plus every future change to them.
5. How often things change afterwards
The quote covers the build. The cost covers the build plus every ERP upgrade, API change, new channel and new rule for as long as you run it. A business that changes its trading rules twice a year has a different total cost from one that never does.
6. Who owns it after go-live
The driver nobody prices. If the integration belongs to the platform supplier, the ERP partner and an agency jointly, it belongs to nobody — and the cost reappears as delay and internal time rather than as an invoice.
One-off, recurring, and the part that hides
Comparing two quotes on their headline figure compares only the first of these three.
One-off — the build
Scoping, mapping, the connection itself, data cleanup, testing and go-live. This is the number on the quote, and it is the part that varies least between competent suppliers.
Recurring — keeping it true
Licence or subscription, monitoring, and the maintenance that absorbs ERP upgrades and API changes. Often quoted as a small percentage, and often the difference between a working integration in year three and a broken one.
Hidden — your own people
Internal time on data cleanup, testing, reconciling what does not match, and the manual workarounds that persist because a flow was never finished. It never appears on an invoice, which is exactly why it is underestimated.
The reason cheap quotes are often the expensive option is that they price the first column accurately and leave the other two with you. We have written that argument out in full in what cheap integration quotes leave out, and the structural version of the same point — why direct connections multiply — is in point-to-point integration risk.
Six questions that turn a quote into a number you can trust
Ask every supplier the same six, including us. The value is less in any single answer than in how precisely each one is answered — precision here is the best available proxy for whether the supplier has done this before.
What happens when the ERP is upgraded?
The answer should describe a process and say who pays for it. "It should be fine" is not an answer.
Who owns this after go-live?
One named party, or three who will point at each other. Ask what happens if the ERP partner and the platform supplier disagree about a fault.
Which of our rules are configuration?
Take your five most awkward trading rules to the call. Anything that needs code is a cost multiplier on every future change.
What does a change cost once we are live?
A new channel, a new price list, a new rule. If there is no answer, changes will be quoted individually at whatever the rate is then.
What is excluded from this quote?
The most informative question on the list. Data cleanup and testing time are the usual exclusions, and they are usually the largest hidden cost.
What happens if you stop trading?
Uncomfortable and worth asking. Whether the integration is documented, transferable and understandable by someone else is a real part of what you are buying.
Our own pricing structure is published on the pricing page, and what we would need to see to turn it into a figure is on when we are the right fit — including the cases where we are not.
ERP integration cost questions
How much does ERP integration cost?
There is no honest single figure, and anyone quoting one without seeing your systems is guessing. The cost is driven by how many systems connect, how many entities move between them, how messy the existing data is, and how often the rules change afterwards. What we can tell you is which of those drivers moves the number most, and what to ask a supplier so their quote means something. That is what this page is for.
Why will nobody publish ERP integration prices?
Partly because the work genuinely varies, and partly because published figures would invite comparison on price rather than scope. There is also a sourcing problem: most of the cost statistics circulating online trace back to vendors who sell the remedy for the problem their number describes, and several trace to reports that are gated, withdrawn or no longer online at all. We checked a number of the widely repeated figures and say plainly on this page which held up.
What makes ERP integration more expensive than expected?
In our experience the recurring causes are data that is messier than anyone admitted at scoping, commercial rules that were described as simple and are not, and the absence of anyone owning the integration after go-live. The first two inflate the build; the third turns a one-off into a permanent drain. Panorama Consulting's 2026 ERP Report names unexpected additional technology requirements as the most common cause of budget overrun, which is the same pattern from the buyer's side.
Is a one-off integration build cheaper than a maintained connector?
It is usually cheaper to buy and more expensive to own. A one-off build is priced against today's requirements and has no budget line for the ERP being upgraded, an API changing, or the person who wrote it leaving. A maintained connector prices that in. Which is better value depends entirely on how long you expect to run it — below about two years the one-off often wins, and beyond that it usually does not.
What should we ask a supplier to get a meaningful quote?
Ask what happens when the ERP is upgraded, who owns the connection after go-live, which of your commercial rules are configuration rather than custom code, what the change process costs once you are live, and what happens to the integration if they stop trading. A supplier who answers those precisely is quoting for the real job. A supplier who answers them vaguely is quoting for the build and leaving you the rest.
Do you publish your own integration pricing?
We publish our pricing structure rather than a single number, because the number depends on the drivers on this page. The structure is on our pricing page, and a discovery call is how we turn it into a figure for your systems.
Want this applied to your systems?
A discovery call works through the six drivers above against what you actually run, and ends with a scope rather than a guess.