2025's Site In Review

(Aside: I'm considering moving this into a microsite powered by Evidence, like I do with my Music In Review, where I could graph some of this data more effectively - we'll see if I get round to it by the end of the year!)

Commentary

As with last year, there's been another significant decline (~20%) in overall hits on my site.

Not that this is a bad thing generally - although I enjoy folks reading my stuff, I don't crave it (well, only a little). I do very much enjoy having folks read what I have to say, and this year I've found that although I've not written as many "viral" posts (which means fewer big spikes of views), I've found more continued readership - including being part of the "Developers Who Write" Bluesky starter pack!

And as noted in previous years:

I noticed that there was a trend downwards [...], but am putting it down to maybe a bit more around folks using things like blocklists for common analytics platforms like Matomo that I use.

We can see a few notable bumps over the year:

• 2024-01-27: I posted Go 1.24's go tool is one of the best additions to the ecosystem in years
  • I expected this to get folks to click more, given a slightly more clickbait-y title, but it was also something I did agree with - albeit I was a bit less enthused with the implementation after I'd really dug into it
  • It did well on Hacker News, reddit, and was then in many Go newsletters - and was the 9th most popular on Golang Weekly over the year
• 2024-02-12: I posted a cheeky related post, Go 1.24's omitzero is another one of the best additions to the ecosystem in years
  • This one didn't get as much discussion going, when actually this may have been a more interesting change
• 2024-08-15: I posted my recap of GopherCon UK 2025
  • This was featured in a few newsletters, and shared by other speakers and attendees alike, so continued on-and-off for a few weeks of views

Although not hugely big in traffic, I had a few other posts of note:

• I'm joining Mend to work on Renovate full-time
• Additional lessons learned running Renovate at scale
• Worries about Open Source in the age of LLMs
• Creating beautiful visualisations of dependency data with Evidence
  • This was a big piece of work leading up to my workshop at GopherCon UK and gave me a lot of insight into building cool things on top of Evidence - there was a lot of work towards this internally at Elastic before I left, and I'm a little gutted I can't continue improving what we were doing there!

And a couple of external posts I authored:

• Starless: How we accidentally vanished our most popular GitHub repos
• Building a more secure npm ecosystem with Mend Renovate

Looking for a new job

Something different about this year has been the looming shadow of needing to look for a job, with me looking to move on from Elastic after some changes happening near me that I wasn't particularly a fan of, and me starting to quietly look around since early February.

I took the opportunity to think about what I actually wanted, wrote up those requirements in a way that was easy to share with others and spent some time also writing about what I'd learned in the previous 2 years (give or take some lessons I couldn't share publicly), as well as continued work on Dependency Management Data and Renovate to feed into what would come next.

Increased scraping

This year I've seen a significant increase in web scraping - I know that LLMs are already trained on a lot of my content, but it seems like a significant increase in scraping - 998 page views (from the same user) in a single visit, but many bots pretending to be real users, but without fingerprinting to show they're the same user, which culminated in a very expensive AWS bill, and I'm still fighting them.

Tweaked LinkedIn crossposting

I tweaked my LinkedIn crossposting from the plain Zapier RSS syndication to a custom setup.

This is partly due to my links no longer presenting as nicely, but also so the posts themselves are a bit more actionable and searchable.

Previously, it would look like:

Whereas now you see:

AI disclaimer

I'm "cautiously skeptic" about LLMs/AI generally, but this year I've definitely been doing a bit more with them - partly as part of entering the job market, but also out of curiosity and seeing what I can do with them, while also being a heavy proponent of the legal/ethical risks that they bring.

This year, I've done a bit more on my blog, publishing content with AI generated content (code or prose). Before I started doing this, I made it possible for me to clarify this in metadata for myself, but also to present to users.

I believe that there should be a disclaimer when using LLM-generated or derived content, and so want to make it very clear.

I hadn't thought that I'd use a lot of it, but looking at the stats, I've got 11 posts that contain some level of LLM-generated content - whether that's the title or code (or content, but that hasn't been the case in the last year).

Single-page RSS feeds

This year I created single-page RSS feeds for my /salary and /now pages, so folks can keep on top of updates to these specific pages.

I also wired up my Salary feed to RSS Parrot so folks on the Fediverse can keep an eye, if so wanted.

Improved experience for keeping an eye on my site's analytics

For years, I've kept one of my monitors showing my site's traffic, as a way of keeping an eye on sudden spikes (i.e. has someone linked to me, or shared my post to Hacker News), but there's some information I've cared more about than Matomo can know.

This year, I built a TUI to give me a more targeted view:

This has been a fun additional side project, getting a chance to play around with Charm's Bubble Tea framework for TUIs, as well as give me much more control over what I want to keep an eye on.

Overall traffic

Traffic across the year

Most popular pages

Most popular posts from 2025

Sources of traffic

Most popular tags