Buyer guide

What ERP actually costs

Not one major mid-market ERP vendor publishes list pricing. That is a category norm rather than an accident, and it means most buyers negotiate without knowing where their quote sits in the distribution. This guide is built from 63 real quotes.

Have a quote to benchmark?

Send it with your requirements. We will tell you where it sits and which lines are soft.

1 / 3
From 63 real quotesInternal time includedImplementation is the variable line

Buyer guide

Five lines in a real ERP budget.

Vendor proposals typically show two of these. The other three are where budgets are missed.

Licence or subscription

The line everyone negotiates. It varies by perhaps 20% between comparable quotes after discounting, and it is the line vendors are least willing to move on beyond a standard band.

Implementation

Median 1.4× first-year licence across our sample, with more than 2× variance for the same scope. This is where the money and the risk actually are.

Integration

Appears in about half the quotes we see, and where it appears it is usually scoped for two systems when the buyer runs eight. That gap surfaces in month four.

Your own team’s time

400 to 1,200 hours depending on entity count and data condition. At loaded cost that is $40,000 to $150,000, and it appears in no vendor proposal we have ever seen.

Year two and three

Seat growth, module additions, and annual uplift mean year three commonly runs 30 to 50% above year one for the same business. Quotes are built around year one.

Negotiate the right line

Most buyers spend their effort on licence. Implementation is quoted by a partner with far more discretion and considerably more variance, and it is a larger number.

The three questions that move an implementation quote in our experience: what was your median variance to original quote last year, what happens if we exceed the estimated hours, and will you name the specific people assigned. Partners who answer all three tend to be the ones worth working with, and the answers themselves change the price.

Licence varies by 20% between comparable quotes. Implementation for the same scope varies by more than 2×. Buyers spend their effort on the wrong one.

Timing

The same configuration quoted in month one of a quarter and in week twelve differs materially. This is well known and still under-exploited by mid-market buyers, largely because internal approval cycles are not planned around it.

The line nobody will quote you

A vendor cannot invoice for your team’s time, so including it makes their number look worse against a competitor who omits it. Every proposal therefore compares licence to licence, and the largest cost on both sides stays invisible.

Estimate it yourself: 400 hours for a light single-entity implementation with clean data, up to 1,200 for multi-entity with data problems. It is the weakest figure in our study — only eleven of sixty-three companies tracked it — and it is still better than omitting it.

Five things that should worry you in a quote

  • A fixed implementation price given before anyone examined your data. That is an opening position in a change-order negotiation, not a fixed price.
  • No exclusions section. The exclusions are the part worth reading. A proposal without them has deferred the disagreement rather than avoided it.
  • Integration scoped for fewer systems than you run. Count them yourself and check the number in the proposal matches.
  • Year-one pricing with no year-three projection. Ask for a three-year total at expected headcount, in writing.
  • A refusal to state historical variance. “Every project is different” means either they do not measure it or the number is bad.

Where our own pricing sits

We publish ours, which is unusual here and partly a positioning decision. Our implementation multiple is roughly 0.25× year-one subscription against a market median of 1.4×.

The honest reason is that we do less: no partner margin, a narrower product, and an integration-first approach that defers the migration. A NetSuite implementation covers more ground than ours does. The number worth comparing is three-year total cost against what you actually need, which is what the calculator is for.

The sample is biased upward

These 63 quotes were shown to us by prospects and customers, usually by buyers who thought the number was high. That skews the sample and we state it rather than correcting for it, because we cannot correct for it honestly. Treat the medians as an upper-middle estimate.

Questions

Common follow-ups.

Why does nobody publish pricing?
Because unpublished pricing lets a vendor charge different customers different amounts for the same configuration. We observe 2.1× variance for comparable scope.
What should we negotiate hardest?
Implementation. It varies by more than 2× for the same scope and the partner quoting it has far more discretion than the vendor has on licence.
How do we estimate our own time?
400 hours for a light single-entity implementation with clean data, up to 1,200 for multi-entity with data problems. It is the weakest figure in our study and still worth including.
Is quarter-end timing real?
Yes, and it is under-exploited by mid-market buyers because internal approval cycles are rarely planned around it.
Why is your implementation cheaper?
Because we do less — no partner margin, a narrower product, and integration first rather than migration first. It is a real difference and not an entirely fair comparison.

Send us the quote.

We will tell you where it sits against comparable deals and which lines have room in them.