I built a simple directory of job boards 4 years ago. It just passed $100K 💵

Rod, thanks for coming back on the attribution question, that detail is more useful than it probably looked.
Bing Webmaster Tools showing it while your analytics does not makes complete sense once you see why. ChatGPT's retrieval leans on Bing's index, so Bing sees the crawl and impression side of it. Your analytics never gets the chance, because a lot of assistant traffic arrives with no referrer at all and quietly lands in direct. Filtering for chatgpt.com and perplexity.ai recovers some of it, never all.
Which effectively makes Bing Webmaster Tools the closest thing to a free AI visibility dashboard right now, and almost nobody has it installed because they wrote Bing off years ago. That is a genuinely good tip and I had not seen anyone say it plainly.
One follow up if you have a second. Did it show up first in the crawl stats or in the search performance report? Trying to work out which panel starts moving earliest, because that determines how small a site can be and still read the signal.

Really like the point about not confusing improving the product with improving the business. A lot of us get tempted to redesign or rebuild when the real value is already in traffic, SEO, data, and community. Curious, out of all your revenue streams, which one became the most reliable over time?

"Don't confuse improving the product with improving the business" is
the line that nails the whole post. Most IH builders spend 90% of their
time polishing features that don't move revenue and call that
"iteration." Your revenue streams (affiliates, marketplace, featured
listings, ads) are sold through curation + SEO + relationships, not
through prettier job board cards.

The deeper unlock is what you didn't build. You didn't build features.
You built supply (boards submitting) and trust signal (organic
listings + community). Both are 100x harder to copy than a redesign,
and neither shows up in a beta launch screenshot.

PHP + raw CSS is honestly the move for a directory. Fancy stacks are
for products where UI is the value. Yours isn't — value is the
inventory and the SEO traffic. Every hour on React would have been
an hour stolen from the work that actually paid.

The moat here, relationships with hundreds of job board founders plus one feed infrastructure powering six channels, never shows up on a pitch deck slide and is nearly impossible for a competitor to buy their way into. I ran an MSP for almost twenty years, and the businesses that sold for real multiples were never the ones with the cleanest codebase, they were the ones with relationships and distribution nobody could replicate quickly. Keep resisting the rewrite, that instinct is correct.

It sounds like you’ve built a really useful tool for job seekers! Personalized job alerts can make a significant difference in how effectively individuals find roles that fit their specific criteria. I’d be curious about how you’re approaching the SEO aspect especially since job boards are often highly competitive in search rankings.

From my experience optimizing content for visibility, focusing on long-tail keywords that align with niche job categories can be valuable. We saw a notable increase in organic traffic when we directed our content strategy towards specific job titles and industries rather than broad terms. This approach often yields better engagement, as your content then attracts a more focused audience.

Also, consider integrating some form of user interaction or feedback, like a simple survey or feedback loop, to understand how well your job alerts and SEO pages are meeting the needs of your users. When we implemented user feedback mechanisms, it not only helped refine our features but significantly improved user retention and engagement by around 30%.

How’s the response been so far to the Telegram bot? I've found that testing different alert formats (frequency, content type) often leads to fascinating insights into user preferences.

The "one piece of infrastructure powering several channels" part really stands out — the Reddit bot, Telegram alerts, and SEO pages all running off the same feeds instead of separate manual work for each. That's a great example of automation compounding instead of just saving time once. Curious — was that multi-channel setup something you architected early on, or did it evolve as you added each new distribution channel one at a time?

Four years to $100K on something this unglamorous is a good reminder that not everything needs to be a rocket ship story. What kept you maintaining it through the slow years, was it steady enough income the whole time, or a long stretch of near nothing before it compounded?

The line I would have expected you to lead with, and you almost buried, is "increasingly, AI citations". You are sat on one of the few sites that can actually watch that happen, because a directory with a page per board is exactly the shape an answer engine reaches for when someone asks which niche job board to use.
It also makes the "don't confuse improving the product with improving the business" point land harder than you probably meant it to. A rewrite would have put the thing that is actually getting cited at risk, and that thing is the structured page per board and the feed data, not the CSS.
Question, since you offered. Did the AI citations turn up in a way you could attribute, or did you only notice once referral traffic started arriving from ChatGPT and Perplexity? Trying to work out how early that signal becomes readable when you are a lot smaller than 800 pages.

The strongest part is that one underlying asset—the founder-submitted feeds—ended up powering search, alerts, SEO, community distribution, and monetization. The fact that the business passed $100K without needing a major redesign is a useful reminder that accumulated distribution can matter more than product polish.

The interesting moat here may be the maintenance loop as much as the directory itself: founder-submitted feeds, recurring sponsors, and multiple channels all depend on the data staying fresh. A lightweight feed-health score—last successful update, stale listings, broken links, and duplicate jobs—could help prioritize outreach and protect trust for both job seekers and board owners. Have you found that data freshness or feed quality becomes the main operational cost as the directory grows?