Case study
Codebase audits that turn into projects
Evaluating prospective clients’ codebases at Designli and turning the findings into proposals a founder can act on.
- 40+
- codebases audited
- ~2 / week
- evaluation pace
- ~15 days
- average time between new clients signed
- 2 wk → 4 d
- proposal delivery time
The job
A company comes to Designli with an existing product and a question: fix it, extend it, or rebuild it? Before anyone can answer, someone has to read the code. That someone is me.
Engagements come in three formats (Impact Week, Solution Lab, and Traction Lab), and the depth of the evaluation matches the format. The codebases range across web, mobile, and backend: JavaScript and TypeScript, Java, .NET, Swift, Kotlin, C++, and Python.
How I run an audit
Evidence first, opinion second. Every finding points to a specific file and line, so nobody has to take my word for it.
- Automated pass. SonarQube static analysis gives a baseline: reliability, security hotspots, duplication, and complexity hotspots.
- Manual architectural review. Tools miss the important parts: how the modules depend on each other, where the data model fights the product, which integrations (payments included) carry real risk, and what will break first under growth. When it matters, I query the data model directly with SQL.
- Translation. The report a CTO reads and the proposal a founder signs are different documents. I frame recommendations around the client’s goals, KPIs, and risk rather than around engineering preferences.
What changed
The reports became a primary driver of new business: on average, about one new client signed every 15 days, as a direct result of delivered proposals. I also automated the evaluation and reporting flow, which cut proposal delivery from two weeks to four days.
What I can’t show
The clients are confidential, so there are no names or screenshots here. What I can talk about is the method, the language mix, and the numbers above.