Score Every Tool by Its 12-Month Cost
What does this tool cost you in month twelve? Not the landing page. The real bill.
Every framework, database, queue, and vendor you adopt has two prices: the one on the pricing page and the one you pay after it's load-bearing. The landing page optimizes for the first week — fast setup, a slick demo, a generous free tier. The pain shows up later, when the thing is wired into production and switching costs are high. The trick is to score the later number before you commit.
The landing page sells the first week
Adoption is front-loaded by design. A getting-started guide gets you to "hello world" in ten minutes, and that early momentum feels like proof the tool is right. But the first week exercises almost none of the surface that will actually hurt: failure modes at 3am, the upgrade treadmill, the parts of the docs that don't exist yet, and the day you need to hire someone who already knows it.
The demo is a best case shown by the people who built it. Your month-twelve reality is the average case run by whoever is on call.
Score the 12-month cost
Before adopting anything non-trivial, estimate its cost across five axes a year out:
| Axis | Ask yourself |
|---|---|
| Operational load | How often does this page someone at night? Who owns it when it breaks? |
| Hiring drag | Can I hire people who know this, or must I train everyone from zero? |
| Licensing | What does the bill look like at 10x today's usage or seats? |
| Vendor lock-in | If this vendor doubles the price or shuts down, how hard is the exit? |
| Scaling cliff | Is there a usage point where this design falls over and forces a rewrite? |
You don't need precise numbers. You need a relative sense of which choices carry a large tail cost and which are cheap to live with.
Act on the score
The decision rule is simple:
- Low 12-month cost: adopt freely, even if it's unfashionable. Boring tech with a shallow tail is a feature.
- High 12-month cost but strong fit: proceed with eyes open. Document the exit plan and the scaling cliff now, while you still have leverage.
- High 12-month cost and weak fit: flag it before you commit. This is the combination that becomes a two-year migration nobody wants to sponsor.
The goal isn't to always pick the cheapest option. It's to never be surprised by a cost you could have seen coming.
Power moves
- Write the exit before the entry. For any lock-in-heavy choice, sketch how you'd leave in one paragraph. If you can't, that's the real price.
- Price the 10x case, not today's case. Free tiers and per-seat pricing look benign until you scale into the paid band. Model the bill at plausible growth.
- Count the humans. Operational and hiring cost dwarf license fees for most teams. A tool only three people can debug is expensive no matter what it costs.
- Prefer reversible decisions. A choice you can undo in a week deserves far less scrutiny than one that takes a quarter. Save your analysis for the sticky ones.
Resources
- Choose Boring Technology — Dan McKinley
- Reversible and irreversible decisions — Amazon shareholder letter (Type 1 / Type 2)
- Total cost of ownership — Wikipedia
- The Cost of Lock-in — Martin Fowler on evolutionary architecture
Shipping AI or cloud systems? Yeda AI audits and hardens production LLM and cloud pipelines. Talk to us · Read the blog