Clouds and counts on the sites

I found a new ad provider that sounds like they'll take on the sites, but they said they couldn't right now because more than half of my traffic was from outside the US. I was like, "Huh?" I have to admit, I'm not engaged like I used to be. I look at stats at the end of the year, and for several, I didn't understand how traffic kept going up, with revenue going down. I never bothered to look at the composition of the traffic. And yes, most of it was from China and Singapore. Yikes.

As I wrote earlier this week, I got Cloudflare in front of the sites, not knowing that it was free. You don't get all of the features, but you do get 10 rules per domain name. (Oh, they also sell domains at cost, so I've transferred most of them over there.) The one rule I created was to block China and Singapore. It's not so much the graph below, but look at the totals for the lines at the top. 147k mitigated requests since Tuesday, which are blocks from my rule, bots and AI scanners. Insane.

Requests aren't just pages, it's all of the images and scripts and things associated with a page, though bots tend to be for the pages specifically. Still, that's a lot of blocking going on. When you look at the page views counted by Google's AdSense, the service that does the ads, you can see what happened when I started blocking on November 2.

Now, it's not easy to compare to the "good old days" when I was paying my mortgage with the sites, making in excess of $2k a month. I found an old spreadsheet that I used to track the revenue and traffic. The problem is that page measurement was never great in those days, especially when using your own logs. Ditto for users. Google Analytics made it better, but early on it still wasn't great. I stopped filling out that spreadsheet in 2015, when I was still totaling about $1,000+ per month (minus Walt's cut for PointBuzz). My best guess is that traffic is half of what it was then. That sounds bad, but half feels like a win when the big platforms (Facebook, Reddit) have taken on most niche communities.

Revenue is the real problem though. Even ten years ago, I had four ad providers chained, where if one lacked an ad, the next would serve it. Now it's just Google. All of those companies are gone. Revenue per thousand pages is often less than a dollar, and that's with ads plastered all over the place in the most obnoxious ways. I don't see the ads, because I'm a "subscriber" on both sites, but I don't know how any one uses it. In fact, that might be part of the problem. I imagine we're chasing people away with that garbage. I'm grateful for the folks paying for it, but I'm mostly subsidizing it.

I have been able to greatly reduce costs. Being in software, yeah, I can't half-ass it. The sites are insanely fast, redundant and never go down, and you don't get that for free. Some price reductions recently in various Azure services has me down to about $250 a month, including what I pay to Elastic, which powers the forum search. And now, the Cloudflare plumbing, for free, makes everything faster, with connections closer to the people using it, and a lot of caching. I love it. Plus, you can't launch a denial of service at it. That's pretty great.

I predict that the AI backlash, as well as mistrust for big platforms, will eventually drive niche communities back to independent sites. People forget that you don't need an app for the web. Well, you do, but it's called Chrome (or Safari, if you insist). Last year I discovered Pinside, and it's awesome. Minimal ads, too. I will happily continue to give them money every year.

What about the sites? I'm about to release one of the biggest lifts in POP Forums since I refactored much of the internals a few years ago. The changelog for v23 is pretty long. (The changes are already on the sites, because I test there.) CoasterBuzz has seen a lot of internal cleanup, especially this week, as problems cropped up from the Cloudflare-ing. After years of having people rate amusement parks, I finally put up the top 25 amusement parks. I have some other ideas too. Meanwhile, Walt says he's ready to start considering "PointBuzz v3," a nod to the fact that we've had v2 for many years. He's handled a lot of the design and data, and I've handles a lot of the coding. I want to re-do photos, and have some idea. So there's some energy there.

It's a weird time, but years of neglect and agentic coding has really made me interested again. I think Walt feels the same. We'll see what happens.