Planning a Custom Business System in Kenya — Without Burning Money

"We need a system" is how most conversations with Kenyan business owners start — usually after the spreadsheet has cracked, the SaaS subscription hit its ceiling, or the third developer this year quoted a number that felt like a ransom. Here's how I help clients decide what to build, what to skip, and what it should cost.

First: do you even need custom?

Honest test: if an off-the-shelf tool covers ~90% of your workflow for KES 3,000–10,000/month, buy the tool. Custom software earns its cost when:

• Your process is your advantage. If how you work is why customers choose you, renting someone else's workflow dilutes it.
• The glue tax is real. When staff spend hours copying data between WhatsApp, sheets and systems, the "cheap" tools aren't cheap.
• You need Kenyan rails natively. M-PESA reconciliation, WhatsApp ordering, SMS fallbacks — foreign SaaS treats these as edge cases; your business treats them as Tuesday.
• Per-seat pricing hurts. 20 staff × KES 2,000/month forever vs one build you own.

Scope the smallest system that proves value

The failure pattern I see most: a "system" spec'd as everything the business does, priced like a small car, abandoned at 60% when the money or patience runs out. The pattern that works:

• Pick the one workflow that bleeds most. Orders? Stock? Collections? Follow-ups? Build that first — completely, end to end.
• Define "working" before "built." "Staff stop copying orders into three places" is a measurable outcome. "A modern ERP" isn't.
• Phase the rest. Reporting, automation and polish land in phases 2–3, funded by the savings phase 1 already produced.

What actually drives the cost

• Workflows and roles — every distinct process and permission level is real work.
• Integrations — M-PESA, SMS, accounting tools, existing databases. Each one has its own failure modes (see my Daraja field guide for how deep that rabbit hole goes).
• Data migration — moving history out of sheets or a dying system is unglamorous and unavoidable.
• Reporting depth — "a dashboard" can mean one chart or a real analytics layer.

Ballpark for a focused first release: internal tools commonly land KES 150,000–400,000; multi-workflow platforms go higher. Anyone quoting without asking about your process first is pricing their guess, not your system.

Five questions to ask any developer before paying

1. "Show me a system you built that's still running." Live production beats screenshots.
2. "What happens to my data and code if you disappear?" The right answer involves your own repo, hosting and docs — not silence.
3. "How do you handle M-PESA callbacks and reconciliation?" If the answer is hand-waving, your payments will be hand-waved too.
4. "What does maintenance cost after launch?" Systems rot without it — get the number in writing.
5. "What would you NOT build for me?" A developer who only says yes is selling hours, not judgement.

This is how I've shipped systems like the Custom CRM (internal ops, real-time, role-based), WritersAdmin (marketplace with escrow) and BookFast (appointment booking). If you're weighing custom vs off-the-shelf — here's how I scope and price it, or describe your process and I'll tell you honestly which side you're on.