Why you should own your link infrastructure
Every link you've shared is a bet on a redirect host staying up. When that host goes away, the bet pays out badly. Here's what owning the layer underneath looks like.
In late 2024, a popular link shortener used by tens of thousands of small businesses pivoted to a different product line and stopped honouring its free tier. Most paid customers kept working. Free-tier customers — and quite a few paid ones whose plans got reclassified — found their links broken with two weeks' notice.
That's the moment people start thinking about who owns their links. Until that moment, links feel like utility — type it, share it, click through. Underneath, every short link is a bet on a third party staying alive and aligned with you. Most of those bets pay off. Until one doesn't.
This post is the version that goes through what "owning your link infrastructure" actually means, the layers that matter, where the failure modes hide, and how much of this is paranoid versus how much is reasonable hygiene for a business that depends on links surviving.
What a link actually depends on
Every short link has four things between the user and the destination:
- The DNS record — pointing the short-link domain at the redirect server.
- The redirect server — the machine handling the 302 or 307 response.
- The redirect rule — the row in the database mapping
/abctohttps://destination.com/.... - The destination — the actual page the user ends up at.
Break any one and the link fails. The destination is yours; you control that. The other three are usually someone else's.
When you use a generic shortener, the platform owns 1, 2, and 3. You're paying them (or relying on their free tier) to keep all three alive forever. When you white-label or self-host, you push more of those layers under your own control.
The four ownership levels
There's a spectrum from "everything is theirs" to "everything is mine". Each step gives you more control and more responsibility.
Level 1: Use a generic shortener as-is
bit.ly/abc, tinyurl.com/foo, t.co/x. They own everything. You own nothing except the destination URL.
Wins: zero setup, zero cost (mostly), works immediately.
Risks: link breaks if the platform goes away, gets bought, changes its terms, or de-prioritises your tier.
Level 2: Use a branded short link on someone else's platform
yourdomain.com/k/abc, where the redirect host is still a third party. You own the DNS record (1) but the platform owns the server (2) and the rule (3).
Wins: brand on the link, escape hatch via DNS (you can point the domain elsewhere if the platform fails — the URL stays the same to anyone who saved it).
Risks: if the platform fails, you have to stand up a replacement service that handles the redirect for the slugs you've been generating. Not trivial, but possible — especially if you've kept a backup of your link records.
Level 3: Self-host the redirect
yourdomain.com/k/abc, with a small server you operate handling the redirect. You own DNS (1), the server (2), and the rule (3).
Wins: full control. No platform to lose. Migrating away is a server migration, not a service migration.
Risks: you have to maintain the server. Uptime, certificate renewal, database backups, monitoring. The redirect server is conceptually trivial but operationally a real thing.
Level 4: Self-host with multi-region redundancy
Same as level 3 but with the redirect distributed across multiple regions, automatic failover, full disaster-recovery posture.
Wins: maximum reliability and lowest latency.
Risks: the work to maintain it is real. Most businesses don't need this level.
What the right level is
Depends on three things:
Volume. A business sending 100 links/month doesn't need level 3. A business that's printed half a million QR codes does — the cost of those breaking dwarfs the cost of running a small server.
Print exposure. A link shared once in an email and never referenced again is low-stakes. A link printed on packaging that ships for the next five years is high-stakes. Print = own the layer underneath.
Risk tolerance. Some businesses (regulated industries, anything with compliance asks) can't afford a third party's outage causing a customer-facing failure. They need at least level 3.
For most small businesses, level 2 is the sweet spot. You get the brand on the link, you keep the escape hatch via DNS, and you don't have to operate a redirect server. If the platform fails, you're inconvenienced but not destroyed.
For agencies and platforms that serve their own customers — where every customer's links would break in a vendor outage — level 3 starts to look attractive even at small scale.
Two-thirds of short-link platforms that existed in 2010 don't exist anymore. Every link generated on the dead ones is dead too.
What level 2 actually buys you
The DNS escape hatch is the part most people don't internalise. Walk through the scenario:
You're at level 1 — bit.ly/abc for everything. The platform announces it's shutting down (or pivoting, or "sunsetting" your tier). Every link you've shared is dead. You can rebuild on a different platform but every printed QR, every old email, every social post is now a 404.
You're at level 2 — yourdomain.com/k/abc on a platform you white-label. Same announcement. You point your DNS at a different platform that supports re-importing your slug-to-destination map. You give them the export. They reconfigure their server to serve your slugs. Your DNS now points at their server. Every old link still resolves, because the URL the user has is still yourdomain.com/k/abc — only the backend changed.
The migration is days of work, not weeks, and it's invisible to anyone who has the link saved. That's worth more than the price difference between a generic shortener and a branded one.
The catch: you only get this escape hatch if the platform supports custom domains in the first place, and you have the slug-to-destination data to migrate. Both are non-negotiable when picking a level-2 platform.
What can go wrong at level 2
Three things to plan for:
The platform exits ungracefully. Tells you with two weeks' notice. You don't have time to fully migrate. Mitigation: keep a regular export of your link records (most platforms support this; some make it a paid feature). The export is your insurance.
The DNS provider has issues. Cloudflare goes down for an hour, your links go down for an hour. Rare but real. Mitigation: a multi-DNS-provider setup is overkill for most businesses; just be aware that DNS is one of your dependencies even at level 2.
The TLS cert fails to renew. The platform's auto-renewal breaks; visitors see a "this site is not secure" warning. Mitigation: monitor the cert expiry yourself with a free uptime tool. The platform should be doing this; you should verify.
None of these are existential. They're each a few hours of recovery, not the death of your link strategy.
What can go wrong at level 3
If you self-host the redirect, the surface area expands:
The server goes down. Memory leak, disk full, OS update, DDoS. You need monitoring, alerts, and a way to restart fast.
The database loses data. Backups, point-in-time recovery, the works. The redirect rule database isn't large but it's catastrophic to lose — every link breaks.
Certificate renewal breaks. Same as level 2 but you're the one running it. Use Let's Encrypt with a tested auto-renewal pipeline.
You run out of capacity unexpectedly. A campaign goes viral, traffic spikes 100x, your server falls over. Mitigation: load-test before campaigns, have a known-good "scale up" runbook.
The work to run this well is real. It's not impossible — a competent infrastructure engineer can stand up a level-3 redirect setup in a week. But the ongoing cost (mostly attention, partly money) is non-zero.
A decision matrix
The pattern: most non-mission-critical link work belongs at level 2. The DNS escape hatch is the right amount of insurance for most businesses, and the operational cost is approximately zero.
What we built around this
Linked.Codes is a level-2 platform. You bring your domain, we run the redirect server and the rule database, you keep your DNS record. You can export everything any time as CSV — that's the migration insurance.
If the platform stops being good for you, you point DNS elsewhere, import the export to the new platform, and your links keep working without anyone noticing. That's the escape hatch the architecture protects.
We also try to behave like a platform that's worth depending on — small operator, no VC pressure to pivot, paying customers from day one, transparent about how the business works. Whether we succeed at being that long-term is a question only time can answer. The architecture protects you regardless.
A simple checklist
If you're using short links for anything that matters, run this:
- Are the links on your own domain? If no, that's the first thing to fix.
- Do you have a recent export of every link? If no, request one or generate one.
- Do you know who maintains the redirect server you're depending on? Is it a business with paying customers and a clear business model?
- If the platform announced shutdown tomorrow with 30 days' notice, do you know what your migration plan is?
If you can't answer those four, you have less link infrastructure than you think you do. Fixing it is mostly a one-afternoon job — point a DNS record, set up exports, evaluate alternatives. The cost of doing it on a calm Tuesday is much lower than the cost of doing it the day after a platform goes dark.
Is using a generic shortener really risky?
For low-stakes use, no. For anything you've printed, depended on for marketing campaigns, or built into a workflow — yes, the risk is real. Platforms have shut down before; they will again.
How often do short-link platforms actually fail?
Major platforms (bit.ly, tinyurl) have been around for 15+ years. Smaller ones come and go more often. The category-leader risk is low; the long-tail risk is significant. The free-tier shutdown risk is the most common pattern.
Can I move from level 1 to level 2 mid-flight?
You can move forward (new links go on your branded domain) but the old generic links stay generic. They keep working as long as the platform does. There's no way to retroactively rebrand existing shared links.
What's the cost of running level 3?
For a small server: $5-20/month for the host, plus the engineer time to set it up (a few days) and maintain it (a few hours per quarter). Cost rises with traffic; for moderate volume it's a small line item.
Should small businesses ever go to level 3?
Rare. Most don't have the volume or the print exposure to justify the operational cost. Level 2 is almost always the right answer. Move to level 3 if your link volume or print exposure makes vendor risk material.
What about decentralised options like ENS or IPFS?
Interesting in principle, immature in practice. The user experience for non-technical readers (browsers don't natively resolve ENS without extensions) limits the audience. For mass marketing in 2026, traditional DNS-based links are the right call.
How do I think about this for QR codes specifically?
QR codes lock you in harder than other links because reprinting is expensive. Putting QR codes on a custom domain (level 2 or 3) is the highest-leverage move you can make to insure them against platform risk. See the post on QR codes vs short links for more on this.
Try it on your own domain
Branded short links and dynamic QR codes, on your subdomain or your own domain. One-time purchase, no per-click fees.
