A good ERP selection starts with knockout criteria, but most knockouts remove less than buyers expect. In 290 runs of the ERP Pilot comparison from 15 February to 19 October 2026, the median buyer still had 16 of 19 systems in play after every hard requirement was applied. Only two answers cut the field sharply: a requirement for on-premise or private hosting, and parallel ledgers for two or more accounting standards. Everything else is decided in the weighted requirements matrix and in scripted demos, which is where the seven phases below spend most of their time.

Most ERP selection processes fail before they even start. Companies invite three vendors to present, watch three demos where everything looks perfect, and choose based on the best sales presentation rather than the best technical fit. Six months into implementation, they discover the system cannot do something critical, and by then switching is no longer an option. Updated 22 October 2026 with a knockout analysis of our own selection data.

💡 Key Takeaway

Settle the two heavy knockouts first (deployment model and parallel ledgers), because each one removes a large block of systems. Then expect a long list of 10 or more candidates and plan the matrix, demos and proof of concept around narrowing that list, not around the knockouts.

Which requirements eliminate the most systems?

Deployment and parallel ledgers, by a wide margin, even though few buyers need them. We re-scored the answers of 290 English selection runs with the current ERP Pilot scoring engine and counted, for each question, how often the answer knocked out at least one system and how many systems it removed when it did.

How often each requirement eliminated at least one ERP system
Manufacturing depth48.4%Industry42.2%On-premise or private hosting27.1%Number of legal entities17.5%Number of countries16.3%Company size16.2%Heavy customisation14.8%Parallel ledgers (2+ standards on one dataset)9.8%Budget per user0.0%
How often each requirement eliminated at least one ERP system
Share of runs
Manufacturing depth48.4%
Industry42.2%
On-premise or private hosting27.1%
Number of legal entities17.5%
Number of countries16.3%
Company size16.2%
Heavy customisation14.8%
Parallel ledgers (2+ standards on one dataset)9.8%
Budget per user0.0%

Source: ERP Pilot quiz, 290 English runs from 15 February to 19 October 2026, re-scored on 22 October 2026 with the current scoring engine. A knockout is a rule in our scoring model, not a vendor statement. Base: runs that answered the question.

The chart shows how often a requirement bites. The table shows how hard. A requirement for on-premise or private hosting bit in 78 of 288 runs (27.1%) and removed 7 systems every time: the ones our model treats as cloud-only (for Dynamics 365 Finance & SCM that is stricter than Microsoft, see our on-premise ERP overview). Parallel ledgers, meaning two or more accounting standards reported from the same postings, were required in only 28 of 287 runs (9.8%), but each time they removed 13 of 19 systems. No other single answer comes close.

Requirement Runs where it bites Systems removed when it bites
Parallel ledgers (2+ standards on one dataset)28 of 287 (9.8%)13.0 of 19
On-premise or private hosting78 of 288 (27.1%)7.0 of 19
Number of legal entities32 of 183 (17.5%)3.3 of 19
Company size47 of 290 (16.2%)3.3 of 19
Manufacturing depth93 of 192 (48.4%)3.0 of 19
Number of countries47 of 289 (16.3%)2.7 of 19
Industry121 of 287 (42.2%)2.2 of 19
Heavy customisation29 of 196 (14.8%)1.0 of 19

Industry and manufacturing depth bite far more often because manufacturers are a large share of buyers: 121 of 290 runs (41.7%) came from a manufacturing sector. Those answers mostly remove the finance and services specialists, two or three systems at a time. Company size, number of countries and number of legal entities remove the small-business systems once a company grows past their range.

Budget is not a knockout

The per-user budget answer did not eliminate a single system in 287 runs, by design. Budget ranges compare against vendor list prices, and list prices are a starting point for negotiation, as our ERP pricing guide shows per system. A system above budget is flagged and loses points, but stays ranked. Put the budget ceiling into the commercial weighting of your matrix, not into the knockout list.

Which ERP systems drop out most often?

The specialists. Sage Intacct and Unit4 were eliminated in roughly two of three runs, mostly because the buyer was a manufacturer. SAP Business One dropped out in 44.8% of runs, usually on complex manufacturing, company size, number of countries or legal entities. At the other end, SAP S/4HANA Cloud Private Edition, SAP S/4HANA On-Premise and IFS Cloud were never knocked out: no answer in our model eliminates them, although several answers cost them points.

Share of 290 selection runs in which each system was eliminated
Sage Intacct69.3%Unit467.2%SAP Business One44.8%Odoo34.8%Dynamics 365 Finance & SCM33.4%QAD33.4%S/4HANA Cloud Public Edition32.1%Oracle Cloud ERP26.9%Oracle NetSuite26.9%Dynamics 365 Business Central19.0%Acumatica19.0%SYSPRO19.0%proALPHA12.8%Epicor Kinetic9.7%Sage X39.7%Infor CloudSuite9.7%S/4HANA Cloud Private Edition0.0%S/4HANA On-Premise0.0%IFS Cloud0.0%
Share of 290 selection runs in which each system was eliminated
Share of runs
Sage Intacct69.3%
Unit467.2%
SAP Business One44.8%
Odoo34.8%
Dynamics 365 Finance & SCM33.4%
QAD33.4%
S/4HANA Cloud Public Edition32.1%
Oracle Cloud ERP26.9%
Oracle NetSuite26.9%
Dynamics 365 Business Central19.0%
Acumatica19.0%
SYSPRO19.0%
proALPHA12.8%
Epicor Kinetic9.7%
Sage X39.7%
Infor CloudSuite9.7%
S/4HANA Cloud Private Edition0.0%
S/4HANA On-Premise0.0%
IFS Cloud0.0%

Source: ERP Pilot quiz, 290 English runs from 15 February to 19 October 2026, re-scored on 22 October 2026 with the current scoring engine. A knockout is a rule in our scoring model, not a vendor statement.

A high elimination rate is not a quality judgement. It says that a product is built for a narrower profile than the average buyer in our data. Oracle NetSuite and Oracle Cloud ERP, for example, dropped out only when the buyer required on-premise or private hosting, in 26.9% of runs. For a cloud-first company, neither ever fell to a knockout.

The resulting long list is longer than most selection guides assume. Only 40 of 290 runs (13.8%) ended with seven systems or fewer. The 9 runs that required both on-premise hosting and parallel ledgers were left with exactly three: the two SAP S/4HANA editions that run on-premise or in a private cloud, and IFS Cloud.

Systems still in the running after every knockout (of 19)
All 19 (nothing eliminated)67 runs15 to 18 systems106 runs10 to 14 systems77 runs3 to 7 systems40 runs
Systems still in the running after every knockout (of 19)
Runs
All 19 (nothing eliminated)67 runs
15 to 18 systems106 runs
10 to 14 systems77 runs
3 to 7 systems40 runs

Source: ERP Pilot quiz, 290 English runs from 15 February to 19 October 2026, re-scored on 22 October 2026 with the current scoring engine. A knockout is a rule in our scoring model, not a vendor statement.

Method and limits

Base: 290 runs of the English ERP Pilot comparison, 15 February to 19 October 2026, after removing internal tests and runs with answer options that have since been retired. Each run was re-scored on 22 October 2026 with the current engine, so older runs count by today's rules. Stage 2 questions were answered in 197 runs. Quiz users select themselves and skew toward manufacturing, and a knockout is our assessment of a hard gap, which you should confirm with the vendor for your exact scope.

Phase 1: Define Your Knockout Criteria (Weeks 1 to 2)

Before looking at any vendor, document your absolute requirements: the things that, if missing, eliminate a system regardless of how good it looks otherwise. Order them by how much they remove. The data above suggests settling deployment and parallel ledgers first.

Typical knockout criteria for mid-market companies include:

  • Deployment model: must be on-premise, must be cloud, or either is fine. Our on-premise ERP overview lists which systems still install on your own servers.
  • Parallel ledgers: multi-GAAP reporting from the same postings (local GAAP, IFRS, US GAAP)
  • Manufacturing complexity: none, discrete, process, or mixed
  • Country localizations: which countries need full legal compliance
  • Integration requirements: systems that cannot be replaced

The same list applies when someone proposes to build the system with AI instead of buying one. Our guide on building your own ERP with AI shows which of these criteria a home-built system fails first.

The power of knockouts

If your company requires multi-GAAP parallel ledgers, you can immediately eliminate Dynamics 365 Business Central, Dynamics 365 Finance & SCM and Odoo, and NetSuite stays in only if its Multi-Book setup passes a test with your auditor. You have narrowed the field in five minutes. That is the power of knockout-first evaluation.

Phase 2: Build Your Requirements Matrix (Weeks 2 to 4)

Beyond knockouts, create a weighted requirements matrix. Because the knockouts typically leave 10 or more candidates, this matrix does most of the narrowing. A workable starting weighting:

  • Functional requirements: what the system must do (50% weight)
  • Technical requirements: integration, performance, customization (20%)
  • Organizational requirements: partner availability, training, support (15%)
  • Commercial requirements: pricing model, contract terms, budget fit, scalability (15%)

Common mistakes

Making the matrix too long (50+ requirements becomes unmanageable). Weighting everything equally (defeats the purpose). Including "nice to have" requirements that dilute the signal.

Phase 3: Long-List to Short-List (Weeks 4 to 6)

Apply knockout criteria first, then score remaining systems against your weighted matrix. Narrow to 2 to 3 finalists.

Typical mid-market short-lists:

Segment Typical Short-List
Enterprise (200+ users)SAP S/4HANA vs. D365 Finance & Operations vs. Oracle Cloud
Mid-Market (50 to 200 users)D365 Business Central vs. NetSuite vs. Odoo
SMB (10 to 50 users)SAP Business One vs. D365 Business Central vs. Odoo

Phase 4: Structured Vendor Demonstrations (Weeks 6 to 10)

Never accept a generic demo. Provide each vendor with a scripted demo scenario based on your actual business processes. Make them demonstrate your specific workflows, not their best features.

A good demo script includes:

  • A realistic order-to-cash scenario using your product types
  • A procurement-to-payment scenario with your approval workflows
  • A month-end close scenario with your reporting requirements
  • An integration scenario showing data flow to and from existing systems

Pro tip

Have your end users attend and evaluate demos. They will spot usability issues that procurement teams miss. The people who use the system daily should have a significant vote in the selection.

Phase 5: Reference Checks and Proof of Concept (Weeks 10 to 14)

Ask each vendor for references from companies in your industry, your size, and ideally your region. When talking to references, ask:

  • What went wrong during implementation (not just what went right)
  • Actual vs. budgeted timeline and cost
  • Post-go-live support quality
  • Whether they would choose the same system again

For your top 1 or 2 finalists, consider a paid proof of concept (PoC) with a fixed scope: your two or three hardest processes, run on your own data. It validates fit before you commit and costs a fraction of a failed implementation, whose cost drivers are set out in ERP implementation costs.

Phase 6: Commercial Negotiation (Weeks 14 to 18)

ERP pricing is always negotiable. Key levers:

  • Timing: quarter-end and year-end are when sales teams have the most room to discount
  • Competition: having two finalists creates leverage
  • Multi-year commits: longer terms unlock better per-user pricing
  • User count guarantees: committing to minimums reduces per-user cost

Contract essentials

Always negotiate implementation costs separately from software licensing. Get a fixed-price or capped implementation proposal with clear scope definitions and change order processes.

Phase 7: Decision and Contract (Weeks 18 to 20)

Your final decision should weight:

  • Technical fit: 40%
  • Total cost of ownership: 25%
  • Implementation partner quality: 20%
  • Strategic alignment and vendor roadmap: 15%

The Faster Alternative

Three inputs make the phases above go faster. For the long list, the best ERP software comparison covers the whole market in one place. For Phase 6, the implementation cost breakdown gives you the benchmarks to negotiate against, and the pricing guide the list prices per system. The architecture question in Phase 1, cloud or on-premise, is worth settling before the requirements matrix rather than after it, because it is the knockout that removes the most systems. Once the contract is signed, the ERP implementation timeline guide shows how long each system typically takes to go live.

The process above takes four to five months. If you want a data-driven starting point in 10 minutes, our free ERP comparison tool applies the same knockout criteria and weighted scoring analysed above to narrow your options. It will not replace a thorough evaluation, but it will tell you where to focus your effort.