1. Establish the real problem
Most searches start with a symptom — slow close, no dimensional reporting, too much manual work. Establish whether that is a ledger problem or a data problem, because the answers cost an order of magnitude apart.
Buyer guide
Nine products regularly reach mid-market shortlists. A typical buyer seriously evaluates two, chosen from brand recognition, and the second is often there to make the first look considered. The middle of the market — where the fit usually is — goes unexamined.
Tell us what you are considering and what started the search. We will give a straight read.
Buyer guide
Buyers under-invest in the second and fourth. Those are the two that predict whether the implementation lands.
Most searches start with a symptom — slow close, no dimensional reporting, too much manual work. Establish whether that is a ledger problem or a data problem, because the answers cost an order of magnitude apart.
Two to three weeks, derived from a quarter of real data rather than from interviews alone. This phase determines everything after it and is the one most often shortened.
Before any vendor is scored. Weights chosen after the demos justify a preference that has already formed.
Three is the number at which scenario demos are affordable and references can actually be called. Document why the others were excluded.
Your transactions, your edge cases, sent in advance. A standard demo tests the vendor’s ability to demo; a scenario demo tests the product.
Ask each vendor for two customers who left or whose projects overran. Some refuse, which is an answer.
A vendor demo is a rehearsed path through clean data. It tells you nothing about how the product handles your intercompany elimination, your revenue allocation, or the customer who pays eleven invoices with one short wire.
Write three to five scenarios from your own transactions, including at least one that is genuinely awkward. Send them to each candidate a week in advance. Attend with your controller — the person who closes the books asks the questions that matter.
Vendor-supplied references are selected, and a call with a happy customer tells you what success looks like rather than what risk looks like.
Ask specifically for two customers whose implementations overran or who subsequently left. Vendors who provide them are usually the ones worth shortlisting; the reluctance itself is data. When you get the call, ask what they wish they had known, what the variance to original quote was, and who specifically was on their project.
Between QuickBooks at roughly $2,000 a year and NetSuite at $150,000-plus there is a large space: Sage Intacct, Acumatica, Dynamics 365 Business Central, Odoo, and integration layers over your existing ledger.
A services business at $40M evaluating QuickBooks against NetSuite is choosing between something it has outgrown and something built for a company four times its size, while four products fitted to exactly its shape go unexamined. Our directory scores ten products on published weights, and we rank fourth on it.
A meaningful share of selection processes should conclude that the existing system is adequate and the real problem is a process, a data quality issue, or one missing integration.
Build that option into the evaluation explicitly, with a cost attached, rather than treating it as a failure of the process. Roughly a third of our own health checks reach that conclusion, and it saves considerably more than any negotiation would.
Most firms marketing themselves as independent ERP selection advisers take referral fees from the vendors they recommend. It is standard, legal, and disclosed rarely and vaguely. Ask the question before the first meeting. We take no fees from any vendor and our selection advisory is paid by the client, which is why we can recommend competitors — and we do, regularly.
Questions
Requirements from data, weights fixed before scoring, and demos run on your own transactions.