Most ERP buyers I meet walk into the decision with a feature matrix. They walk out with a contract. They almost never walk in with a real ROI model.
That is how companies end up paying for software that delivers nothing measurable. Here is the framework I use — four numbers, one spreadsheet, no consultants required.
Number 1 — current operating cost of the stack you are replacing
Add up everything. Licences. Integration platforms. The analyst whose job is mostly data reconciliation. The two finance hires whose week is mostly chasing AP entries. The Excel models nobody can edit. The external consultant on retainer for reports.
For a typical 100-person company we audit, this number is 2.2× to 3.5× what the licences alone suggest.
Number 2 — working-capital improvement
When inventory accuracy moves from 88% to 98%, when DSO drops from 72 days to 54, when stockouts halve — cash is released. Quantify this conservatively. Talk to your auditor.
For most mid-market businesses, working-capital improvements alone pay for the platform within 18 months. They never appear in vendor decks. They should.
Number 3 — revenue lift from speed and intelligence
This is the hardest to estimate and usually the largest. Faster lead response. Better cross-sell. Lower churn. Pricing optimised on real demand instead of last year's gut feel.
A reasonable approach: pick the three commercial KPIs the platform claims to move, look at what comparable customers achieved, take the lowest number and halve it. If the deal still works on that number, it is a real deal.
Number 4 — implementation cost honestly stated
Licence cost is the small number. Implementation cost is the big one — and the place vendors quietly hide assumptions. Force a fixed-price implementation with clear scope, or build a multiplier of 1.5× into the variable quote.
The 12-month threshold
Add #1 and #2. Subtract licence cost and #4. If the result is positive within 12 months — without depending on #3 — buy. The revenue upside is the bonus, not the case.
If the result is positive only when you bake in #3, slow down. Push the vendor for a structured pilot with success-tied pricing. Most modern platforms — including ours — will do this for serious mid-market buyers.
When the math says wait
It is fine to conclude "not yet." If your stack is small and your operations are simple, the integration tax is not yet large enough to justify migrating. The right answer for a 10-person services firm is almost never a new ERP.
But once you cross the 50-person, multi-channel, multi-entity threshold, the math almost always flips. The question stops being whether to consolidate — it becomes how soon you can.