Key Takeaways
- Start from the three numbers you cannot get today, not from a feature list. The right ERP is the one that produces them with the least friction.
- You are choosing between four options — a subscription product, an industry-specific suite, open source you host, or a custom build — and each wins in a specific situation described below.
- Total cost over five years at your future headcount, who owns the data, and what happens when your process changes are the three questions that separate good choices from expensive ones.
To choose ERP software well, work backwards from the problem: write down the two or three operational numbers you cannot get today (real margin per job, stock cover per SKU, on-time delivery rate), map the workflow that would produce them, and only then compare options against that map. Businesses that start from vendor feature lists end up buying the longest list; businesses that start from their own workflow end up buying the system they will actually use. The rest of this guide is that process, in order.
Step 1: Define the Problem in Three Numbers
Before any demo, finish these sentences:
- "Every month I wish I knew ______ and today it takes ______ hours of spreadsheet work to find out."
- "Our team re-types ______ from ______ into ______, and it goes wrong about ______ times a month."
- "We lost or delayed ______ last quarter because we did not know ______ in time."
Those answers are your requirements. A system that fixes all three is a success even if it lacks 200 features on someone's comparison chart. Common answers we hear: margin per job or product, stock on hand by location, order status without calling the warehouse, month-end close time, and quotes that reflect real cost.
Step 2: Map the Workflow That Produces Them
Take one order — a real one from last month — and write every step from the first customer contact to the money arriving in the bank, with who does it and which tool they use. Do the same for one purchase and, if you make things, one production run. Mark every step where something is re-typed, waited for or guessed.
That map is the single most useful document in the whole selection. It tells you which modules you need first, where integrations must exist, and — crucially — how standard or unusual your process really is. Standard processes fit products; unusual processes fight them.
Step 3: Know the Four Options and When Each Wins
| Option | Wins when | Watch for | Cost pattern (US, SMB) |
|---|---|---|---|
| Subscription (SaaS) ERP | Your processes are standard, your team is small, you want to start in weeks | Per-seat fees rising with headcount; configuration limits; vendor roadmap decides features | US$100–$250 per user per month plus implementation of US$15k–$100k+ |
| Industry-specific suite | You are in a vertical with a mature suite (e.g., distribution, construction) and your process matches the industry norm | Expensive to leave; customization by consultants; features aimed at bigger companies | Custom-quoted; first-year totals often six figures |
| Open source, self-hosted | You have in-house technical staff who will own it, and license cost matters more than time | You become the vendor: hosting, upgrades, security, and the gaps are yours to fill | No license; real cost is engineering time and hosting |
| Custom build | Your process is genuinely non-standard, per-seat costs are climbing, or you want the system to change as fast as the business does | Choose a partner that phases delivery and hands you the code; avoid big-bang upfront projects | Upfront: high five to six figures; or flat monthly — TechAuditPros US$1,800/mo all-in |
Notice that "custom" and "cloud" are not opposites. A custom ERP is normally cloud-hosted in an account you own — see What is cloud ERP? for the distinction.
Step 4: Ask Every Vendor (Including Us) These 15 Questions
The questions that separate good options from expensive ones
- Show me how my order (from the map) flows through your system, step by step. Not a demo — my order.
- Which of my steps require a workaround, a customization or a third-party add-on?
- What is the total cost over five years at the headcount I expect in year three?
- What exactly is included in implementation, and what is billed separately?
- How long until the first module is in daily use? How long until all of them are?
- How do you migrate my existing data, and how many trial migrations are included?
- Which of my current tools do you integrate with natively, and which need middleware?
- Who owns the data, where is it hosted, and how do I export everything if I leave?
- Who owns the code (for custom builds), and can another team maintain it?
- What happens when my process changes next year — who makes the change and what does it cost?
- Who exactly will I talk to after go-live? A named person or a support queue?
- What are your backup, recovery and access-control practices?
- Is there a minimum contract term, and what are the exit terms?
- Can I talk to a current customer of my size and industry?
- What will you tell me I do not need?
The last question is the tell. A vendor who cannot name a module you should skip is selling a list, not solving a problem.
Step 5: Watch for These Red Flags
- Pricing only after a discovery call. Ranges exist for every option; a vendor who will not give any is optimizing for negotiation, not fit.
- "Yes, with customization" three or more times. You are about to pay for a product and a custom build. Pick one.
- A twelve-month implementation for a forty-person company. Something is oversized — the product, the scope or the partner.
- Demo data that is never yours. If they will not load a sample of your real products and orders before you sign, they are not confident it fits.
- Long contracts with exit fees. Confidence in a product does not need a lock-in.
- No answer to "who owns the data?" Walk away.
Step 6: Decide, Then Phase
Whichever option you choose, insist on phasing: one module, one department or one site first, with the rest following once it is in daily use. It de-risks the decision, produces value in weeks instead of quarters, and gives you a real exit if the fit turns out to be wrong. Our guide to ERP implementation covers how to run that rollout.
If your workflow map shows a process that no product quite fits, that is the case for a custom build — and it is exactly what TechAuditPros does for US businesses and Canadian businesses on a flat monthly rate with the first module live in 6–10 weeks. If the map shows a standard process and a small team, we will tell you a subscription product is the better buy. Either way, the map decides.
Frequently Asked Questions
What is the best ERP software for a small business?
The one that produces the numbers you cannot get today with the least friction and the lowest five-year cost at your future headcount. For standard processes and small teams that is often a subscription product; for non-standard processes, growing headcount or heavy integration needs, a custom build usually costs less over five years and fits better.
How long should ERP selection take?
For a small or mid-size business, two to four weeks is plenty: a week to write the three-numbers brief and workflow map, two weeks of demos run against your own order and data, and a decision. Selection projects that run for months usually mean the problem was never defined.
Custom ERP or off-the-shelf: how do I decide?
Count the workarounds. If demos of off-the-shelf systems fit your workflow map with one or two minor workarounds, buy the product. If you hear "with customization, yes" three or more times, your process is non-standard and a custom build will fit better and usually cost less over time.
What questions should I ask an ERP vendor?
Run your own real order through their system; ask what needs workarounds; get the five-year cost at future headcount; clarify what implementation includes; confirm data and code ownership and exit terms; ask who you will talk to after go-live; and ask what they would tell you not to buy.
Want a second opinion before you buy software?
Book a free strategy call. We look at how your business actually runs and tell you honestly which system fits — CRM, ERP, both, or neither yet.
Book a Free Strategy Call