The Trizlink story: solving the link fragmentation problem

Link fragmentation is what happens when the links you publish are managed by several tools that do not know about each other. Each one issues its own URLs, counts its own events with its own definition, and reports in its own dashboard — so the total is not a total, it is three numbers you are expected to add up yourself. Trizlink started as a complaint about exactly that.

The complaint

A small team runs a campaign. The ad goes out with a shortened link. The same offer sits on a link-in-bio page because that is where the social traffic lands. A few posts go out through a scheduler. Two weeks later somebody asks which of the three actually worked, and the honest answer is that nobody can tell you without a spreadsheet.

It is not that any of the tools is bad. The shortener counts clicks accurately. The bio host counts page views accurately. The scheduler counts impressions accurately. The problem is that a click, a view and an impression are three different events, measured by three systems that have never been introduced, and the campaign was one thing.

The moment that pushed it

The specific moment was a reconciliation that took most of an afternoon and produced a number I did not believe. Three exports, three date formats, two different ideas of what a unique visitor is, and a bio page whose outbound links were not tracked at all because they had been pasted in as plain URLs. The number at the bottom of that spreadsheet was a guess wearing a suit.

The fix was not a better spreadsheet. It was noticing that all three surfaces end in a link somebody clicks, and that if one system issued every one of those links, the reconciliation would not need to happen at all.

What that turned into

Trizlink runs short links, link-in-bio pages and social publishing out of a single workspace. A bio page is built from Trizlink short links rather than pasted URLs. A scheduled post shortens its links automatically. So every click on every surface arrives with the same definition, in the same workspace, on the same timeline, and the report is one report.

That is the whole architecture, and it is deliberately unexciting. There is no clever attribution model here. There is one system issuing the links, which removes the need for a model in the first place.

What we are stubborn about

• Saying what a number is not. Click analytics are best-effort everywhere. Bots inflate them, privacy blockers remove real clicks, and chat-app link previews add clicks nobody made. We report detected bot traffic separately instead of folding it into your total.
• Not storing visitor IP addresses. They are not collected, so they are not held and cannot leak — which is also why location is coarse and labelled as an estimate rather than presented as fact.
• Limits written down before you pay. Every plan lists what it includes; there is no tier whose contents you discover after the card is charged.
• Export on every plan, including Free. Your links and your analytics leave as CSV whenever you want them to.
• One well-implemented sign-in rather than three half-implemented ones — which also means there is no Trizlink password to steal.

The rebuild, and why it happened

The version you are reading this on is a rebuild. The first Trizlink worked, and it accumulated the thing every fast-moving product accumulates: features that existed in the type system and not in the interface, screens that described capabilities the code had never implemented, and documentation that had drifted far enough from the software to be actively misleading.

Rather than patch that, the whole product was audited feature by feature and rebuilt against a written record of what actually shipped versus what was merely claimed. Every public URL was frozen first, so no address anybody had saved could be lost in the process, and every screen was designed and approved before a line of the new app was written.

The visible consequence is the tone of these pages. When something does not work — the browser extension, a social platform rejecting a post, the limits of a click number — it says so, because the rebuild started from a list of places where the old product had not.

Small on purpose

Trizlink is built and maintained by one developer. That is why the prices look the way they do, why there is no "contact sales" button anywhere, and why the roadmap is short. It is also why the honesty above is not a marketing angle: there is nobody here to write around, and a claim that turns out to be false is a claim I have to answer for personally.

If you want to see whether the through-line argument holds for your own work, the Free plan is the cheapest way to find out, and the plans page lists every limit before you commit to anything.