The Signal
A post hit the Hacker News front page arguing that most engineering organizations operate without any real economic visibility into their own output. They track velocity , story points, and headcount — but almost never answer the basic question: what does it actually cost to ship this feature, and what does it return ? The post, by Viktor Cessan , frames this as a systemic blindness baked into how large teams are structured . Managers optimize for proxies. Proxies drift from reality. Nobody notices until the burn rate becomes a crisis. It 's a well-worn critique, but it landed with 100+ upvotes and 50 comments because it's still true in 2025.
Builder's Take
Here's what's interesting if you're building alone or in a tiny team: you don't have this problem. You are the economic unit. You feel every wasted hour directly . There's no abstraction layer between your time and your output.
But most solo builders don't exploit this advantage — they just accidentally avoid the worst failure modes. That's leaving leverage on the table.
The Cost/Capability Curve for One-Person Shops
Naval's framing applies hard here: code is infinite leverage. A solo dev with the right stack can output more working software per dollar than a 20-person team flying blind . But only if you're deliberate about it.
Think about it in first principles:
- A 10-person eng team at $150K average fully-loaded cost = $1 .5M/year. If 40% of that time is coordination, meetings , and rework (conservative), you're getting ~$900K of actual output.
- A solo builder with $500 /month in AI tooling + $200/month in infra can ship a functional SaaS. Real marginal cost per feature: hours of your time + pennies of compute .
- The leverage ratio isn't 10x. It can be 100x — if you measure and optimize it.
The trap Cessan identifies in large orgs — optimizing for proxies instead of outcomes — is also available to solo builders who get lazy. Tracking GitHub commits instead of revenue. Measuring features shipped instead of retention . Don't do that.
What Moat Does This Create or Destroy?
This is actually a moat creation story for indie builders. Large orgs can't easily fix their economic blindness — it's structural. Layers of management exist precisely because measuring individual output is hard at scale. You don't have that problem. You can know , with a spreadsheet and one hour of work, exactly what your cost-per-feature is and what each feature contributes to MRR.
That visibility compounds. You cut the losers faster. You double down on what works. You're running a tighter feedback loop than any enterprise eng team can, by definition.
Tools & Stack
If you want to actually operationalize economic visibility for your one-person build, here's a concrete stack:
Time Tracking (Cost Side )
- Toggl Track — free tier works for solo. Tag time by feature/project. Export to CSV weekly .
- Clockify — free, unlimited projects. Better reporting than Toggl free tier.
- Both integrate with Zapier/Make if you want to pipe data somewhere.
Revenue Attribution (Output Side)
- Stripe Dashboard — if you're using Stripe, you already have MRR/feature correlation data. You just need to tag it.
- PostHog — open source, free up to 1M events/month. Track feature usage → correlate with retention. Self-host on a $5 H etzner VPS or use their cloud.
- Plausible — privacy-first analytics, $9/month. Simpler than PostHog if you just need page-level data.
The Gl ue: A Simple Economic Scorecard
You don't need a dashboard . You need a habit. Try this in a Google Sheet or Notion table:
| Week | Feature | Hours | Infra Cost | MRR Delta | Cost /$ Return |
|------|---------|-------|------------|-----------|---------------|
| W1 | Auth | 6h | $0.40 | +$0 | ∞ ( foundation)|
| W2 | Export | 4h | $0.10 | +$120 | ~$0 .27/$1 |
| W3 | Dark mode | 8h | $0.00 | +$0 | 💀 cut it |
Your time has a cost. Even if you don 't pay yourself yet, assign a number — $ 50/hr, $100/hr, whatever your opportunity cost is. Now every feature has an honest ROI.
AI Tooling That Changes the Denominator
- Cursor / Windsurf — both ~ $20/month. Compress your hours-per-feature number significantly for bo ilerplate and integration work.
- Claude API / GP T-4o API — check current pricing at anthropic.com and openai.com. Use for automating the analysis layer: pipe your time tracking export into an LLM prompt weekly to get a plain-English breakdown of where your time went.
Ship It This Week
Build a personal engineering economics dashboard — a dead-simple tool that answers: " What did I ship this week, what did it cost me, and what did it earn? "
Here's the minimal version you can start today:
- Set up Toggl or Clockify — create one project per product, tag entries by feature name. Takes 10 minutes.
- Pull Stripe MRR weekly — Stripe's dashboard shows weekly MRR change. Screenshot it or use their API.
- Write a simple prompt — every Friday, export your time log CSV and paste it into Claude or GPT-4o with this prompt:
You are analyzing my solo dev time log for the week .
Here is my time data: [ paste CSV]
My MRR changed by: [+/- $ X]
My infra costs this week: [$X]
My hourly opportunity cost: [$X/ hr]
Give me:
1. Cost per feature shipped
2. Which feature had the best ROI
3. Which feature I should kill or pause
4. One sentence on where I should focus next week
That's it. You just built an economic feedback loop that most $10M ARR companies don't have. Run it every Friday for a month and watch how your prioritization sh arpens.
The large orgs Cessan writes about will still be flying blind next year. You don't have to be.