System Development: Buy vs. Build for a Small Retailer
A regional retailer with 12 stores wants a point-of-sale and stock system. The in-house IT team is two people who currently maintain the network and email. A packaged SaaS product costs 300,000 yen per store per year, includes support and updates, and covers 90% of requirements; the missing 10% (a loyalty-points feature) could be handled manually at first. A custom build is estimated at 40 million yen up front plus 6 million yen per year for maintenance, and would take 14 months with the two-person team plus contractors. The retailer's annual revenue is 900 million yen.
1. Compare the two options over five years on cost and risk.
Read the worked answer
Five-year cost:
- SaaS: 300,000 × 12 stores × 5 = 18 million yen, all-in, with support and updates included.
- Custom: 40 million up front + 6 million × 5 = 70 million yen, before cost overruns.
Cost strongly favors SaaS (18M vs 70M). Risk also favors SaaS: the two-person team cannot realistically deliver and then maintain a 14-month custom system while keeping the network running. A custom build concentrates key-person risk in a tiny team. The 10% gap (loyalty points) is a manual workaround initially and can be revisited. The only scenario favoring custom is if the missing 10% were core to competitiveness - it is not, per the description.
2. The owner says 'we already have IT staff, so building is cheaper.' How do you correct this reasoning?
Read the worked answer
The reasoning confuses sunk salary with zero marginal cost. The two staff are already paid to run the network and email; assigning them to a 14-month build removes that capacity and adds contractor cost (already in the 40M). Their salaries are not free capacity.
Also, the comparison must include maintenance, which the owner omitted: 6M/year forever, versus SaaS's bundled updates. And it must include opportunity cost - 14 months without the system means 14 months of manual stock work across 12 stores.
The correct frame: compare total five-year cost and risk, not the sticker price of development labor. On that frame, SaaS wins decisively.
3. What conditions would make you reverse the recommendation toward custom build?
Read the worked answer
I would reverse if any of these held:
- The missing 10% were a core differentiator (e.g., a loyalty program central to the business model) that no package could supply.
- The retailer had a real in-house engineering capability, not two network admins.
- Data-residency or integration constraints forbade SaaS.
- Scale made per-store SaaS pricing exceed build cost - e.g., hundreds of stores, where 300,000 × N grows without bound while custom cost is roughly fixed.
None of these hold at 12 stores and 900M revenue. Stating the reversal conditions shows the recommendation is reasoned, not reflexive.