I’ve had this tab about “A big list of things I disable in WordPress” sitting open for…longer than I want to admit.

Here’s the post about where I actually try it. And, well. Wasn’t worth the wait.

(This is part of the “how do I webcomic?” series, with useful information on all kinds of comicking-related topics.)

Testing ground 1: Shine

Started on And Shine Heaven Now, because that site has the lowest stakes if it breaks something. Which…I did. Because I didn’t double-check the formatting of my functions.php file.

Once I had the correct opening/closing tags in there, though, everything seemed fine! For comparison, I ran this Shine landing page through PageSpeed Insights, both before and after the WP-disabling.

The mobile diagnostics seemed like they had an immediate, dramatic boost. From a score of 51 to 71. (Both “mid” by the site’s color-coding, but 50 is the very bottom of the “mid” range.)

…And then I did the exact same test 10 minutes later, and it was back to 51.

Specific data points PageSpeed used to calculate that score:

• First Contentful Paint: 6.5 s (bad) – 2.3s (mid) – 6.4s (bad)
• Largest Contentful Paint: 12.8 s – 9.0 s – 11.9s (all bad)
• Total Blocking Time: 40 ms – 80 ms – 70ms (all good)
• Cumulative Layout Shift: 0.171 (mid) – 0.034 (good) – 0.171 (mid)
• Speed Index: 6.6 s (bad) – 3.6s (mid) – 6.5s (bad)

Desktop diagnostics were in the green the whole time, with all the numbers extremely close. Score went from 94 to 96 to 97 (the third test is when the mobile results plunged again, but the desktop results didn’t even dip). Again, the numbers:

• First Contentful Paint: 0.6 s (every time)
• Largest Contentful Paint: 1.2 s – 1.2s – 1.1s (for some reason the 1.2 is “good” the first time, and “mid” the second)
• Total Blocking Time: 150 ms (mid) – 90ms (good) – 80 ms (good)
• Cumulative Layout Shift: 0.003 (every time)
• Speed Index: 0.9 s – 1.2s – 0.8s (all good)

I was curious enough to wait another 10 minutes and do a fourth test, but this one errored out. Even with that much of a delay, my web host apparently flagged it as Suspicious Bot Traffic, and rate-limited it.

So, yeah, very possible that did nothing at all?

It wasn’t a total waste of time, though — PageSpeed also flagged a couple insecure HTTP links in the layout. So at least I updated those to HTTPS.

Testing ground 2: BICP

I still wanted to test-drive all this on another site. One with slightly more active users, now that I was confident I at least wouldn’t crash it for somebody in the middle of reading.

So I ran PageSpeed’s analysis on But I’m A Cat Person’s final “comic” update.

BICP did much better on mobile from the start!…And then plunged in the first test with all the WP-disabling PHP in place.

This time I waited a solid 15 minutes before doing a second test. This time, it was consistently bad both times!

Overall score went from 82 to 67, and was still 67 when I re-checked:

• First Contentful Paint: 2.1 s (mid) – 3.5s (bad) – 3.5s (bad)
• Largest Contentful Paint: 4.4 s – 6.8s – 6.8s (all bad)
• Total Blocking Time: 0 ms (every time)
• Cumulative Layout Shift: 0.026 – 0 – 0 (all good)
• Speed Index: 3.2 s (good) – 5.0s (mid) – 5.1s (mid)

The desktop numbers were good, and stayed good. Overall score went from 99 to 98 to 99 — it’s barely worth comparing, everything was in green, and I’m sure these differences are all within a normal margin of error:

• First Contentful Paint: 0.5 s – 0.6s – 0.6s
• Largest Contentful Paint: 0.9 s – 1.0s – 0.9s
• Total Blocking Time: 0 ms (every time)
• Cumulative Layout Shift: 0.008 – 0.007 – 0.007
• Speed Index: 0.8 s – 1.0s – 0.6s

There is no testing ground 3

PageSpeed had its own suggestions for “ways to improve this page’s performance.” Maybe some other day I’ll investigate those. But whatever other hangups my sites have, this de-bloating effort wasn’t helping them.

I was really hoping this would be a nice uplifting post about One Weird Trick that you, too, can use to make your webcomic site faster.

Instead, it’s an “I tried this so you don’t have to” post. Oh well. Can’t win ’em all.