ERP software comparison: test the close, not the modules
The useful ERP software comparison for a South African business is not a module list. It is two numbers from each vendor's reference customers: how many working days the month-end close takes, and how many figures are still re-typed between systems along the way.
Every package on a shortlist has a ledger, purchasing, stock and invoicing. The difference you will feel, and pay for, is whether the management pack is ready on day three or day twelve, and how much of it a person assembled by hand.
Why the module list misleads
A comparison grid with ticks in every row tells you each vendor can do the job. It does not tell you how the job gets done in your business. A tick next to "inventory" says nothing about whether the stock count from your warehouse app arrives in the ledger or gets typed in on the 28th. Mid-market packages such as SAP Business One are sold through partners, so what you usually see is the partner's demo data, not your close.
The a16z essay You Need a New CFO puts the problem plainly. Finance teams spent "most of their energy at the bottom of the pyramid stitching together inputs". They were pulling exports, reconciling spreadsheets and chasing variances "buried in an email thread". That is a description of a slow close, and no module list shows it.
The month-end close test
Ask every shortlisted vendor for one reference customer of your size, in your sector, in South Africa. Then ask that customer, not the salesperson, five questions:
- On which working day is the management pack signed off?
- Which numbers in it come from outside the ERP?
- How do those numbers get in: an integration, an import, or typing?
- Who does the bank reconciliation, and how many days does it take?
- What did the last change to a report cost, and who made it?
The answers give you a close time in days and a count of manual hand-offs. Those two numbers are your comparison.
What a fast close looks like now
The a16z piece claims the newest tools have moved the target. Rillet, an ERP it lists among its portfolio companies, "turns a monthly close into a daily one". OpenAI's finance function, under CFO Sarah Friar, is described as building toward "a zero-day close". One of the AI skills built by Anthropic's finance team is said to produce a monthly financial review that is "90-95% complete" in about thirty minutes.
Read those as an investor's account, not audited results. None of them is a South African reference customer. What they do show is direction: the close is becoming the measure buyers compare on.
For a South African business of 40 to 200 people, a realistic test is simpler. Can the pack be signed off within five working days, with no number re-typed? If the answer is no on the current system, ask why before assuming a new one will fix it.
What slows a close down, and where it lives
Look for the delay in the joins before you look inside the ERP:
- Bank feeds that arrive late or need manual matching.
- Payroll run in a separate package and posted as a journal.
- Stock counts captured on paper or in a warehouse app.
- Job costs held on job cards that never reach the ledger.
- Intercompany charges agreed by email.
Each join is a place where a person exports, reformats and types, and each one adds time to the close.
That changes what you compare. A package that ships with a bank feed for your bank, a payroll link, and an open interface for your job cards will close faster than one with more modules and fewer joins. The a16z essay also says the new generation of tools "stands up in days rather than quarters". Ask any vendor that says so for the reference.
Putting a rand figure on the gap
Count the hours your finance team spends on the close each month, and multiply by a loaded rate. At an illustrative R250 an hour, a close that takes three people four eight-hour days costs about R24 000 a month. If a package halves it, that is R144 000 a year, and a number you can set against the licence quote.
Then ask what each package costs to change. The joins will move as your business grows, and the cost of adding the next one belongs in the comparison.
When the answer is not a new ERP
If your close is slow and the delay sits in the joins, replacing the ledger can leave every join in place. The faster route is often to keep the ERP and fix the hand-offs into it, one at a time, starting with the one that costs the most days.
That is the work Zorah does: an audit to find the slowest join, one workflow fixed properly, then the rest of the systems you already run connected. Our SAP Business One comparison shows where replacing a package is still the right call. On Monday, ask your bookkeeper on which working day last month's pack was finished, and which three numbers in it were typed in by hand.