The hardest code to write is the code that solves a problem nobody has. We've shipped features that took 6 weeks to build and got 0 user adoption because the underlying assumption was wrong. A design sprint is the cheapest way to avoid that mistake — 5 days, 8 users, and a clickable prototype is enough to kill a bad feature before you commit a sprint to it.
The 5-day structure
We run a variant of the Google Ventures sprint, tuned for B2B SaaS:
- Monday: Map — align the team on the problem, the user, the goal
- Tuesday: Sketch — every participant sketches solutions individually, then critiques
- Wednesday: Decide — pick a solution, build a storyboard for the prototype
- Thursday: Prototype — build a clickable Figma prototype (no production code)
- Friday: Test — 5–8 user interviews, watch them use the prototype, synthesize
The fintech case study
Our client — a B2B fintech — wanted to build a 'smart insights' feature for their dashboard. The hypothesis: users would value auto-generated insights on top of their financial data. Estimated effort: 6 weeks of engineering.
We ran a 5-day sprint. By Friday afternoon, we had tested the prototype with 8 users. 6 of 8 didn't open the insights panel. The 2 who did, didn't act on the insights. The hypothesis was wrong. We killed the feature.
“5 days of sprint cost saved 6 weeks of engineering. That's a 30x return on time. The math is brutal — there is no scenario where skipping the sprint is the right call.”
When NOT to run a sprint
Sprints aren't free — they take 5 days of senior attention from product, design, engineering, and (often) the client. They pay off when the cost of being wrong is high. They don't pay off when the feature is small, the hypothesis is well-tested, or the team has strong existing signal.
Our rule: if the engineering estimate is more than 2 sprints, we run a sprint. If it's less, we ship and measure.
Author
Sofia Renaud
VP, Engineering, Globantis Labs
Sofia writes about engineering at Globantis Labs. Reach out if you'd like to discuss this article or your own project.
