How to choose an ERP for a small business (a scorecard, not a demo)

Most small businesses choose an ERP the same way: they watch three demos, feel most impressed by one of them, and sign. Eighteen months later, roughly half of those businesses are running spreadsheets alongside the ERP they are still paying for.

The demo is a terrible selection instrument. It is designed and rehearsed by people whose job is to make the software look inevitable, using clean data and a happy path. Your business will supply neither. What follows is the method I use instead.

What should you do before looking at any ERP?

Before you look at a single platform, document how your business actually runs. Not how the SOP says it runs — how it runs at 6pm on the last day of the month when someone is chasing a dispatch.

For each core flow — procure-to-pay, order-to-cash, inventory movement, production if you manufacture, month-end close — write down every step, who performs it, what system or paper it touches, and where it habitually breaks. This takes one to two weeks and it is the highest-leverage work in the entire project.

It matters for two reasons. First, you cannot evaluate process fit without knowing your process. Second, and more importantly: an ERP encodes whatever process you give it. If your process is chaotic, you will pay a large sum of money to automate chaos, and it will be harder to fix afterwards than it was before.

Which factors actually predict ERP success?

Weight these against each other, score every candidate out of ten, and multiply.

1. Process fit (weight: 35%)

Take the five processes you documented and walk each one through the platform — ideally in a trial instance with your own data, not a vendor sandbox. Score how much of each flow the standard product handles without customisation.

Be ruthless about the difference between configuration (settings, fields, workflows the product is designed to accommodate) and customisation (code that must be maintained and may break on upgrade). A platform needing heavy customisation for your core flow is a platform fighting your business.

2. Five-year total cost of ownership (weight: 25%)

Licence cost is the number vendors compete on and the smallest part of the total. Model all of it over five years:

• Licences or subscriptions, at your projected user count in year five, not year one
• Implementation and data migration
• Customisation, plus re-testing that customisation at every major upgrade
• Hosting and infrastructure, if self-hosted
• Annual support and maintenance
• Internal staff time — the largest hidden cost, and the one nobody budgets

Open-source platforms like ERPNext shift cost from licences to implementation and maintenance. Proprietary suites do the reverse. Neither is inherently cheaper; the shape of the cost differs, and which shape suits you depends on whether you have internal technical capability.

3. Your team's ability to maintain it (weight: 25%)

This is the factor that gets skipped and then decides the outcome. Ask honestly:

• Is there anyone internally who can own this system after go-live?
• How many implementers exist in your city or country for this platform? What happens if your partner disappears?
• How steep is the learning curve for your least technical user — not your most technical one?

A platform your team can operate at 70% capability beats one they can operate at 20%, even if the second is objectively more powerful.

4. Data portability and exit cost (weight: 15%)

Before you enter a system, know how you would leave it. Can you export complete, relational data — not flattened CSVs of individual screens? Is there a documented API? Is your data stored in a database you can read directly?

You are unlikely to migrate. But the cost of leaving is what determines your negotiating position at every renewal for the next decade.

Which ERP usually wins the scorecard?

Run this honestly and a few patterns show up repeatedly:

ERPNext tends to win for businesses with real process complexity, cost sensitivity and either internal technical capability or a trusted implementation partner. Low licence cost, deeply customisable, genuine open source. It demands more implementation discipline than its price suggests.

Odoo tends to win where breadth of modules and interface quality matter, and where the business wants a large app ecosystem. Watch the per-user, per-app pricing as headcount grows — model it at year-five user counts.

Zoho tends to win for sales-led businesses already inside that ecosystem, where CRM is the centre of gravity and manufacturing or complex inventory is not.

Custom development wins only when your core operating process is genuinely unlike anything packaged software models, and that process is a competitive advantage. This is rarer than founders believe. If your differentiator is your product or your service, buy the ERP and spend the engineering budget on the differentiator.

What should you ask an ERP vendor?

In every vendor conversation, ask:

1. "Show me a client of my size and industry who went live, and let me speak to them unaccompanied." Hesitation here is disqualifying.
2. "Which parts of my process will this product not handle well?" Every product has weaknesses. A vendor who claims none has either not listened or is not being straight with you.
3. "What does year two cost, assuming we grow 40%?" This surfaces per-user pricing cliffs and support renewal jumps before you sign.

Decide, then commit

Once the scorecard has a winner, stop evaluating. Prolonged selection is a common failure pattern — businesses spend nine months comparing platforms when the implementation discipline, not the platform choice, was always going to determine the outcome.

A competently implemented second-best platform beats a badly implemented best one, every single time. Choose deliberately, then put your energy into the implementation.