Weekly Traffic Across 65 Sites: technologies.org LCP Jumped 431% in One Week
Pulled the weekly numbers for July 26 to August 1 and the top line is boring, which is fine. 15,630 visits across 65 sites, down 3.58%. Page views 16,130, down 4.1%. Median load time 3,252ms, basically unchanged. Nothing on fire at the aggregate level.
The interesting stuff is always one layer down though.
technologies.org is the thing I actually need to look at. LCP went to 4,252ms at the 75th percentile. Up 431% in a week. That’s deep in the “poor” band, which starts at 4 seconds. Visits also dropped 31.76% but I don’t think those two are connected, at least not yet. Core Web Vitals damage doesn’t hit rankings inside seven days. Still, a five-fold jump in LCP isn’t noise. Something got added. Plugin update, an image without dimensions on it, a script in the header, cache quietly dying. CLS on the same site actually got better (0.15, down 25.84%), which tells me it’s not a template change. More likely one fat asset. So the job is archaeology: figure out what landed on that site around the 26th.
analysis.org went the other way and I’ll take it. Visits down 20.25%, sure, but LCP down 37% to 2,816ms and CLS down 40.6% to 0.21. INP at 56ms. That’s the previous round of fixes showing up in the data. Not done though. 0.21 CLS is still twice the threshold I want, and 4,831ms median load is the slowest of everything that reports properly, even after a 31% improvement.
osint.org is the one site I have zero complaints about. Up 25.42% on visits, LCP at 2,135ms which is comfortably in the green, CLS down to 0.13. Median load time did jump 66% to 3,656ms and I noticed that, but LCP is what search actually scores, so I’m not going to chase it this week.
Now the part that bugs me a bit. Look at the browser splits on the top two sites. analysis.org: 720 of 1,300 visits logged as Unknown, versus 390 Chrome and 50 Mobile Safari. technologies.org: 520 Unknown against 610 Chrome. Unknown at that ratio is crawlers. So the real human number on both is roughly half what the dashboard says, and any 20 or 30 percent weekly swing on those two might just be a bot deciding to visit more or less often. Worth checking the Chrome plus Mobile Safari subtotal before reading anything into a move.
And then there’s the blind spot. Sites four through ten report no load time and no Core Web Vitals whatsoever. Not bad numbers. No numbers. Visits are getting counted at the edge but nothing is running client side, which means the beacon script isn’t in those templates. That’s marketanalysis.com, pho.tography.org, transportational.com, opinion.org, mediapresser.com, appcoding.com and technologyconference.com. Together that’s 3,760 visits a week, about a quarter of the network, that I have no performance visibility into at all. One snippet in the base template fixes it. No excuse for this one.
Movers on the good side: appcoding.com up 91.3%, mediapresser.com up 66.67%, marketanalysis.com up 64.44%. The first two are new builds so that’s just indexing catching up to what got published. Not demand, not yet.
Bad side, besides technologies.org: pho.tography.org down 30.34% and transportational.com down 24.29%. Both dropped far enough that something probably fell off page one. Need to check what.
Three things for this week. Find whatever broke on technologies.org. Get the beacon onto those seven sites so I’m not guessing. Finish the CLS work on analysis.org while it’s still responding to attention.