Â鶹Éç

Everyone told me directories were dead.

Google had supposedly killed them. Social media had replaced them. And now AI was going to finish off whatever was left. I heard some version of this for years, usually from people who had never actually run one.

I launched in 2007. For years, rebuilding it was one of those ideas that lived permanently on the back burner — I knew the legacy codebase had a ceiling, I just never found the moment to act on it. That moment finally came in February 2026. By then, the site had accumulated 30,000+ verified listings, 55,000 referring domains and nearly two decades of editorial decisions baked into every category, every slug, every approved submission. Walking away from that felt less like a business decision and more like burning down something I had quietly devoted two decades of my life to.

So I didn’t walk away. I spent the next two-and-a-half months working eight to 10 hours every single day to rebuild it. What came next nearly broke me anyway.

The migration nobody warns you about

The original DirJournal ran on a legacy PHP directory script (phpLD). It worked, in the way that a 1987 pickup truck works — reliably, until the day it doesn’t, and then all at once. Moving 30,000-plus listings to a modern stack without breaking URLs, losing data or destroying nearly two decades of accumulated SEO equity is the kind of problem that sounds manageable until you are three months in and realise you have barely started.

We initially hosted the rebuild on Netlify. It was the obvious choice at the time, but we quickly ran into a wall: Netlify’s prerendering support did not work reliably for a site of our scale and structure. Search engines were not seeing what users were seeing, which for a directory built on the premise of being found, was a fundamental problem. We moved the entire stack to Vercel, which meant reconfiguring deployments mid-rebuild while everything else was already in motion.

Redirects alone consumed weeks. We had 7,731 redirect rules spread across three separate files, each one mapping a legacy URL to its new equivalent without dropping a single page that had ever earned a backlink. On top of that came 1,050 blog stub posts that needed individual audit decisions — which ones to flip live, which ones to redirect to a consolidated page, which ones to retire entirely. Category slug consistency had to be standardised across the entire site. Every decision had a downstream consequence. Fix one thing, break two others.

There was a specific week, somewhere in the middle of all this, where I genuinely considered stopping. Not pivoting, not pausing. Stopping. I was staring at a 404 error report that was 40 pages long with no clear sense of how far down the problem went. The technical debt was enormous, the revenue during the rebuild was not, and I could not yet see the finished version clearly enough to know if it would be worth it. I sat with that feeling for a few days. Then I looked at the backlink profile we had built since 2007 and thought about what it would cost someone to replicate it from zero. The answer was: You cannot. Not in any reasonable timeframe. That was enough to keep going.

The data work nobody sees

The visible rebuild is the one people notice. The invisible one is where the real time went.

Every one of the 30,000-plus listings needed schema enrichment across more than 30 schema types, each one assigned based on business category rather than applied as a blanket template. A law firm gets different structured markup than a SaaS company or a digital agency, because they are different entities and search engines need to understand them differently. We went through four separate rounds of country and state field corrections as we found inconsistencies in the legacy data. Historical payment records spanning thousands of transactions needed reconciliation to restore accurate tier information to listings that had been misclassified for years. CollectionPage schema had to be verified and corrected across category pages to ensure the structured data held up under scrutiny.

None of this shows up in a product launch post. All of it determines whether the site holds up when it matters.

What we actually built

The rebuilt DirJournal is not a directory in the way most people picture when they hear that word. It is closer to an entity verification hub.

Every listing goes through human editorial review before it goes live. We verify the business, check the details and assign structured schema markup to each entry. This level of structured, category-specific markup is what helps businesses qualify for Google’s Knowledge Graph, an increasingly important signal as search results become more answer-driven and less link-driven. This is not automated. It cannot be automated without losing the thing that makes it worth anything.

We also found ourselves needing to build tools that reflected where search and discovery were actually heading. The first was a that shows businesses how their listing performs across trust, expertise and authority signals. Then came an AEO Checker to evaluate how well a business is positioned to appear in answer-based search results. Then an to surface whether a business is being cited when people ask questions in their category. None of these were on the original roadmap. All of them became necessary once we understood what the rebuild was really for.

Why human curation is the moat

Here is what the last year taught me about directories in a world saturated with generated content. The problem is no longer finding information. The problem is trusting it.

When a business appears in DirJournal, it has been looked at by a human being who made a judgment call. That judgment, repeated 30,000 times over nearly two decades, creates something that a scraper cannot replicate and a competitor starting fresh cannot buy. Trust compounds. It accrues slowly, invisibly, and then one day it is the thing that sets you apart from every directory that spun up last year on a template.

SEO professionals understand this intuitively because they have watched Google’s quality signals evolve. The sites that survived every major algorithm update were not the ones that optimised most aggressively. They were the ones who built something worth trusting. The same logic applies here.

The content glut did not make DirJournal irrelevant. It made the case for what we were already doing stronger than I could have made it myself.

What I would tell anyone sitting on a legacy asset

Do not confuse age with obsolescence. They are not the same thing. A domain with nearly two decades of genuine editorial history, real referring domains and a loyal user base is not a liability. It is infrastructure that would take years to rebuild and that most people starting today will never have.

The question worth asking is not whether your legacy asset is old. It is whether it is doing the right job. If the answer is no, the answer is almost never to shut it down. It is to figure out what job it should be doing now and rebuild accordingly.

That rebuild will be harder than you expect. The migration will take longer. There will be a week where you want to stop. Do not stop.

Nearly two decades of trust is not something you walk away from. It is something you earn the right to carry forward.