How to Shortlist ERP Vendors Without Getting Sold a Pitch Deck

A 60-employee specialty food distributor ran an ERP RFP in 2025 and invited five vendors to present: SAP, Oracle NetSuite, Microsoft Dynamics 365 Business Central, Odoo, and Acumatica. Three of those five were never realistic candidates for a company that size, and the RFP process spent six weeks finding that out the hard way, sitting through demos built for a very different buyer, then negotiating pricing that assumed an implementation scope the company was never going to need. A shortlist built around actual tiers, not a wish list of recognizable names, would have cut that process to two weeks.

The three real tiers, and why cross-shopping them wastes time

ERP vendors do not compete evenly across company sizes, no matter how their sales teams present it. There are, broadly, three tiers, and matching a company to the right one before the RFP goes out saves most of the wasted effort in a vendor selection process.

The lines blur at the edges. A fast-growing 80-employee company with three warehouses and a complex fulfillment operation might genuinely need NetSuite's depth rather than Odoo's simplicity, and a $600M company with a lean, single-entity structure sometimes runs Dynamics 365 comfortably instead of stepping up to SAP. But inviting SAP to present to a 60-person distributor, or inviting Odoo to a 2,000-employee manufacturer with operations in six countries, produces a demo that looks impressive and answers questions nobody asked. Filtering the shortlist by tier before sending an RFP is the single highest-leverage step in the process, and it is the step most companies skip because tier-appropriate vendors feel less exciting to evaluate than the recognizable enterprise names.

Implementation cost, not just license cost, generally scales with tier too. A mid-market Dynamics 365 or NetSuite rollout for a company in the 100 to 300 user range typically runs 1 to 1.5 times the first year's license cost in implementation services. Enterprise SAP or Oracle rollouts commonly run 3 to 5 times the first year's license cost, sometimes more, once integration, data migration, and change management are priced in. A vendor's sales quote almost never leads with this number, which is one more reason a like-for-like total cost comparison across finalists matters more than comparing license quotes alone.

Build an RFP that forces proof, not description

A standard RFP template asks vendors to describe their capabilities. Every vendor's answer to "does your system support multi-currency consolidation" is yes, because the honest answer is almost always technically true and completely unhelpful. A better RFP asks vendors to demonstrate specific scenarios using something close to the buyer's actual data and workflow, not a generic demo script the sales engineer has run two hundred times before.

Concrete techniques that work:

• Send a real dataset. A sanitized export of 200 to 500 rows from the current system, such as a chart of accounts, a sample of transactions, or a product catalog with the actual complexity of kitting, lot tracking, or multi-unit-of-measure the business deals with, and ask the vendor to show it configured in their system during the demo, not a canned dataset from their standard script.
• Ask for the three hardest workflows first. Every business has two or three processes that are genuinely awkward, such as a consignment inventory arrangement, a revenue recognition schedule tied to milestone billing, or a multi-entity intercompany elimination. Put those at the top of the demo agenda instead of the end, where they often get rushed or waved off with "we can build that in configuration."
• Require a named implementation team, not a sales team, for at least one call. The people selling the deal and the people implementing it are frequently different individuals from different parts of the organization, sometimes a different company entirely if a reseller or systems integrator is involved. Insist on meeting the actual consultants who would run the project before signing, not just the account executive.

Questions that expose a weak implementation partner

For mid-market and small-business tiers especially, the software vendor and the implementation partner are often separate entities. NetSuite and Dynamics 365 are both sold through a network of resellers and systems integrators, and the quality gap between partners implementing the identical software is enormous. A weak partner can turn a good product into a failed project. A few questions reliably separate strong partners from weak ones:

• "How many go-lives has your team completed in the last 12 months, and how many of those went live on the original planned date?" A partner that cannot answer the second half of that question, or answers vaguely, has not tracked its own delivery performance, which is itself a signal.
• "What was your most recent project that went over budget or over schedule, and what caused it?" Every implementation partner has had a rough project. One that claims otherwise is not being straight with you. The specificity and honesty of the answer matters more than the fact that a rough project happened.
• "Who is the named project manager and lead consultant for our engagement, specifically, not generically?" Get names before signing, not after. Partner firms sometimes staff the sales-cycle demo with senior people and the actual implementation with a more junior bench.
• "What does your post-go-live support look like after the standard hypercare window ends?" A partner with a vague answer here is planning to hand the account to a generic support queue the moment the project closes.

Reference checks that go beyond the vendor's list

Vendors provide reference customers who will say positive things; that is the nature of a curated reference list, not a red flag by itself. The value is in the follow-up questions a curated reference is less prepared for:

• "What would you do differently if you started this implementation again?" This question routinely gets a more honest answer than "would you recommend this vendor," because it does not ask the reference to contradict their own positive framing.
• "Did the go-live date you were originally quoted match the actual go-live date?" A pattern of slipped dates across multiple references, even when each individual reference explains away their own delay, is worth noticing.
• Ask for one reference the vendor did not select. LinkedIn searches for the vendor's implementation partner tag, plus a direct outreach to a company in a similar industry and size range that is not on the provided list, produce noticeably more candid answers than the curated set.

How long this should actually take

A thorough vendor shortlist and RFP process for a mid-market company runs four to six weeks end to end, not the three to four months some organizations let it drift into. A rough breakdown that keeps momentum:

• Week 1: Tier filtering and an initial vendor list, typically three to four vendors, not six or eight. More vendors past that point adds coordination overhead without meaningfully improving the decision.
• Weeks 2 to 3: RFP distribution, the real-dataset demo, and the hardest-workflow walkthroughs described above.
• Week 4: Reference calls and implementation partner vetting, run in parallel with initial pricing negotiation rather than after a vendor is already selected.
• Weeks 5 to 6: Final scoring, contract review, and a decision.

Processes that stretch past eight weeks tend to lose momentum inside the buying organization rather than gain rigor. Stakeholders start missing demo sessions, the original requirements drift as the team's day-to-day priorities shift, and vendors sense the deal cooling, which shows up as less flexibility on pricing by the time a decision finally gets made.

Total cost is part of this evaluation too, and it is worth modeling before a final decision rather than after the contract is signed. The TCO calculator can help put implementation, licensing, and ongoing support costs from different vendor quotes on the same basis, since the sticker price on the software license is rarely where the real cost difference between finalists shows up.

The distributor that ran the five-vendor RFP eventually chose Dynamics 365 Business Central, the same choice a tier-appropriate shortlist would have pointed to in week one. The six weeks spent on SAP and Oracle demos were not wasted for the wrong reasons; the team learned things about pricing structure and implementation scope that informed the eventual negotiation. But a tighter shortlist from the start would have reached the same decision with a fraction of the internal time spent.