Sometimes, when I’m feeling nostalgic, I like to kick back with a 20-oz. bottle of Surge soda and reminisce about the early days of the internet, when search engine optimization was as simple as stuffing the words “Everclear fan site” in your footer 10,000 times. (Sorry, BTS fans, I’m old school.)
Things are more complicated today, and Google has gotten smarter about detecting SEO practices that are targeted purely at bots, rather than humans. You can’t stuff keywords anymore, and you can’t selectively serve some content to Googlebot and other content to real users.
But, Google has created a new frontier in pleasing the bots in the form of Core Web Vitals, the algorithm for measuring the speed of a site that was made part of the ranking algorithm last year. The basic idea is that if your site is faster, you will rank higher in search results.
But as with any algorithm (or the SAT!) part of it reflects reality, and part of it is a game. And recently, WP Tavern reported that WP-Optimize, a plugin owned by the team behind the popular Updraft plugin, included code that seemed like it sort-of maybe could have been trying to game the system by changing the code on the site only for the tools that measure PageSpeed, but not for actual visitors.
In a later article, released a few hours after the original publication of this post, WP Tavern’s team expanded on the details of the problem while WP-Optimize’s team stated that the code in question had no malicious intent and wasn’t built to trick bots or deceive users. I see no evidence that there was intentional deception here from the current WP-Optimize team, but the question of serving different code to PageSpeed bots is worth exploring, since it sheds light on some of the weird incentives created by Core Web Vitals.
[midroll_1]
Over at HDC, we are pros at optimizing for CWV. (I recognize the irony of linking to myself in a post about SEO, but I am shamelessly doing it anyway.) Even so, we recognize that CWV is largely about giving sites that already invest a lot in search-engine optimization an edge over their competition. This is why, in my view, the first 0-70 points on PageSpeed reflect “real” improvements that people might actually notice, whereas the last 30 points are mostly under-the-hood stuff that only bots pick up. As a result, CWV is not like accessibility and ADA compliance – which are all about real people. Instead, it’s about some legit speed improvements and a lot of stuff that no layperson would really notice.
There is a ton of value in getting that last 20-30 points on CWV if you are building a business on SEO traffic. But the same could be said of adding “in 2022” to all your headlines. A lot of this stuff is just Google’s preference, rather than anything that objectively helps people load a site faster or find the right product.
So I think it’s important not to conflate CWV optimization with “what your real users are experiencing” – yes, some users will notice a marginally faster site, but a lot of it truly is under the hood and meaningless to end users! And there are all sorts of ways that CWV recommendations could actually make a site less usable – for example, by requiring extra clicks to view YouTube videos rather than embedding them, or by discouraging nicer fonts in exchange for system fonts that don’t require files to load. CWV even complains about Google’s own tools, like Analytics, being on the site. Some of it is just learning the game.
What we learned from this bizarre episode is that Core Web Vitals scores are “just a number” that matters to clients – but many users simply won’t notice a difference between a site scoring 60 and a site scoring 99. In other words, you can conceivably “cheat” because the only way to tell the difference between a site with a “yellow light” and a site with a “green light” is via these complex scanner tools.
There’s value to being really good at satisfying the scanners – but that’s not always the same as improving the site for your users.
This article was updated on August 30 to reflect the second WP Tavern article and the denial of wrongdoing by the WP-Optimize team, both of which were published after this article’s initial release.
