We turn down projects. Not often, but regularly. Usually it is a business that wants to build a custom system to replace something that an off-the-shelf tool already does perfectly well. Or a startup that wants to build an internal tool before they have validated that anyone will pay for their product. Or a company that wants to add AI to something where the existing manual process is fast, accurate, and costs $200 a month. We turn these down because building them would not help the business — it would consume resources that the business needs for things that actually matter.
The one-sentence ROI test
Before committing to any significant technology investment, we ask the client to complete this sentence: "We are building this because it will [specific outcome] by [measurable amount] within [timeframe]." If they cannot complete it, the project is not ready. This is not a paperwork exercise — it is the forcing function that separates technology that earns its keep from technology that feels important in a meeting.
"We are building this because it will reduce order processing time from 4 minutes to 30 seconds, saving 20 hours per week within 3 months." That is a real ROI. "We are building this to improve operational efficiency" is not.
The true cost of ownership
Every system you build has a maintenance cost. Someone needs to keep it running, update its dependencies, handle the edge cases that the original spec did not anticipate, and train new team members on how it works. This cost is almost always underestimated in the initial project, and it compounds every year the system exists. We tell every client to double whatever maintenance estimate we give them — and to ask themselves whether the system will still be worth its cost in three years, not just in the first three months.
When not building is the right answer
We recommend not building when:
- An existing tool solves 90% of the problem for less than the cost of one sprint.
- The process the system would automate is still changing frequently.
- The team that will maintain the system does not yet exist.
- The expected users are fewer than 10 people and the volume is low.
- The ROI depends on assumptions that have not been validated.
None of this means technology is not valuable. It means technology is most valuable when it is chosen deliberately, scoped tightly, and measured honestly. The businesses we have seen get the most out of their technology investments are not the ones that build the most — they are the ones that build the right things at the right time and maintain them well.