What to fix before you buy software
Software does not fix a process, it accelerates one. If your records are inconsistent today, a system will make them inconsistent faster and more expensively. Six things worth sorting first — most of them free.
Every figure below is hypothetical. The pattern is not.
We build business systems for a living, so this is an odd thing for us to publish. But the most common reason those systems fail has nothing to do with the software, and it is better said in advance than discovered in month three.
Software does not fix a process. It executes one faster. Point it at a clean process and you get speed and visibility. Point it at an inconsistent one and you get inconsistency at higher volume, with an audit trail proving it, and a licence fee.
Here is what to check first. Most of it costs nothing.
1. One place where each fact lives
The commonest condition in a Nigerian SME is not no records. It is several records that disagree: sales in a notebook, invoices in Excel, payments in a bank app, stock in a storekeeper's head, and customer promises in WhatsApp.
Before any system, decide for each fact — a sale, a payment, a stock movement — which single place is authoritative. Not where it is recorded; where it is true.
If you cannot answer that today, a new system just becomes a sixth place, and the first question in every meeting stays "which number are we using?"
2. Records made when the thing happens
A record written at the moment of the event is data. One reconstructed on Friday is a memory, and memories are wrong in predictable directions — see the hours you never invoiced, where the direction is consistently under.
Test yourself honestly: if a delivery went out at 4pm today, when does it get written down? If the answer is "when the driver returns" or "at the end of the week", fix that first. Same-day recording is a habit, it is free, and no software can install it for you.
3. Numbering that holds
Invoices, delivery notes and receipts each need one unbroken sequence. Every system you might buy assumes this, and migrating a broken history into a new system is where implementations quietly lose two weeks. Full detail in invoice numbering that survives an audit.
4. A customer list that is actually a list
Not "Chidi", "Chidi Motors", "Chidi M." and "chidi (new number)" as four entries.
De-duplicate before you migrate. Every duplicate becomes two customers in the new system, with the balance split between them, and it will not be spotted until somebody is chased for money they have already paid. One afternoon now saves that phone call.
5. Terms that exist in writing
Payment terms, credit limits and prices should be recorded per customer, not carried in the owner's head.
This is the one that most often blocks an implementation, because it turns out the terms are the owner's head — different for each customer, remembered rather than written, and occasionally different from what the customer believes. Writing them down surfaces disagreements you would rather find now than during a go-live. It is also the precondition for anything in should you sell diesel on credit.
6. Somebody whose job it is
Every system needs one person who owns the data. Not the most senior person — the one who will notice on Tuesday that Monday's deliveries were not entered.
Systems do not fail loudly. They fail by drifting: entry slips a day, then three, then people start checking the old spreadsheet because it is more current, and within two months the software is a place data goes to be ignored. One owner, checking daily, prevents all of it.
The honest test
Ask yourself: if we hired a new person tomorrow, could they run a day of operations from our written records alone?
If yes, you are ready, and software will multiply what already works.
If no — if the answer requires somebody's memory, somebody's phone, or "you'd have to ask Tunde" — then that gap is your actual problem. Software will not close it. It will digitise the parts that are written down and leave the rest exactly where it is, except now you are paying monthly.
What this means for buying
None of the six needs a vendor, a budget or a decision. They need a fortnight of deliberate habit change, and they make every subsequent option cheaper — including doing nothing.
If you fix these and decide you do not need software after all, that is a completely legitimate outcome, and you are better off than you were.
If you fix them and still want a system, you will implement it in weeks rather than months, because the hard part will already be done. The hard part was never the software.