A URL shortener looks like a weekend toy until a team starts depending on it. Then the job stops being “make a short string” and becomes something quieter: a thin ops layer for aliases, stable redirects, and click signals you can trust. That is the bar I used on the URL shortener case, and on the live tool at tools.danieldiaz.dev/shortener. No vanity scale claims. Just the parts that earn their keep.

What actually matters

Aliases people can remember. A short opaque hash is fine for a paste. An alias you can say out loud is what ops, sales, and support end up using in chats and docs. If the alias layer is sloppy, the shortener becomes another place to look up a link. Redirects that stay boring. The contract is simple: this path still goes there tomorrow. Broken redirects train people to keep the long URL in a spreadsheet “just in case.” Stable redirects are infrastructure. Clever ones are a liability. Click signals that stay honest. Counts only help if they mean something you would defend in a review. Inflated totals, mystery bots, or vanity dashboards teach the wrong lesson. Better a small honest number than a theater metric.

Why it belongs with tools, not with company product

ECORPA carries hiring timelines and third-party portals. A shortener should not inherit that weight. It lives on tools because the blast radius is smaller and the promise is narrower: same craft, different contract. That split is why tools lives apart from ECORPA.

The kill condition

I would turn the shortener off (or stop recommending it) if aliases drift into chaos, redirects break without a loud fail, or click stats become something I would not show a teammate without disclaimers. That is the same falsifier shape as Falsifiable automation: write the observation that kills the claim before you celebrate the demo. A short link layer earns the word “infrastructure” when it removes friction from real workflows. It loses that word the moment you start selling scale you did not earn.