Google Lighthouse and the Perils at Sea
It is perfectly comprehensible that we all want our own websites on page one of search results for certain keywords in every search engine there is — especially with Google, the quasi–standard when it comes to find and being found on the Internet. It also makes some sense that the speed at which our sites may be loaded by a browser and their overall performance are taken into account, but don’t lose your mind over it.
Search Engine Optimization Is Important
Yes, Search Engine Optimization (SEO) is important when you want your online documents to be ranked high (or at all) by search engines, and it is a bit difficult to show up in search results without having been found, and indexed, and ranked reasonably well first. I guess there cannot be two minds about that.
Yet … SEO is not a science — at least not by traditional definition. A cynic would probably state that it is equal parts art and voodoo. I would not go as far as that, but I do find it difficult to completely and utterly deny the pinch of truth in that notion.
After all, where does the abundance of perfectly irrelevant search results come from, despite a great many SEO tools available and already implemented? And where are the many excellent websites, when we search for relevant keywords? And, last but not least, why is it that websites advocating certain practices do not also implement them? It’s like seeing a Vidal Sassoon model run about with a miserable hairdo — each and every day.
Speed Tests Are Important
I already implied that search engines — particularly Google — consider (and honour) your website’s speed. To this end, a number of parameters, considered important for good user experience, are measured. That’s fine and dandy, and increasingly important. After all, we don’t have all day to wait for some content to eventually show and, perhaps, proof its worth to us.
Yes, speed matters, but it should not be — and, in fact, is not — the be–all and end–all of great user experience. And it most certainly is not the “make or break” criterion for good ranking. Come to think of it, nothing is, really. Unless you seriously messed up your code, chances are your site will appear on page one upon someone’s query — whether they were actually looking for content like yours or not.
As I said, SEO is not a science, and those speed tests and the insights you may (or may not) gain from them are not scientifically relevant — and, perhaps even more important, they are no golden ticket for page one, or two, or three …
If you measure an entity several times in short succession under equal conditions (same environment, same procedure, etc.) you may expect to get equal results — with minute variations. That’s called “repeatability”.
Speed Tests Are Not Testing for Actual Speed
Consequently, if you take Google’s performance test several times, checking the same website against their algorithm, you will have to allow for slight deviations. Apparently, your site is tested with a variety of (emulations of) end user devices, even during the same session (that is, without closing or resetting your browser). That’s fine, too (and their business altogether).
Actually, you should not even recognise the differences, because the test is supposed to show the speed of your website rather than the speed of any given end user device. Otherwise, we would speak about “reproducibility” (as opposed to the above mentioned “repeatability”).
Just leave me alone with your silly hair splitting, you may think. After all, different users do use different devices. That’s true, but then your site should be tested against a great variety of available devices for usability (or whatever), but not for loading speed.
Google’s “page speed insights” are supposed to tell you how you could possibly reduce obstacles for your website’s good performance “in general”. In other words, what practices to introduce and which ones to better avoid to deliver a website that provides a reasonable user experience in the shortest possible time. That’s what the developer and designer may influence. Therefore, testing the same website twice in short succession should return identical results.
Short of relapsing into the bad habit of “optimising for this or that environment”, there is no way of telling (let alone influencing) which device any one visitor is going to use to consume your website. I consider it safe to assume that’s nothing any of us wants. We’ve been there already and it was not lovely.
So why even put your website to the test? Naturally, I have no proof, but in general I would assume the whole point of “page speed insights” is to remind developers that there is always more than one way to every goal and more than one (type of) user to consider — and, of course, to avoid big blunders.
Is a page scoring high more likely to rank better? I’d say, “it depends” (on a variety of other factors, and on who is running it). If your scores influenced your ranking really that much, a lot of rather well–known websites would never show in your search results — including some famed for their overall design and actual ranking (not to mention the testing page itself).
Don’t Worry (too much) About Mixed Signals
At times, you may find yourself on the receiving end of quite confusing messages. Especially so, as compliance with web standards is repeatedly mentioned as a decisive factor for good ranking. The problem is, at least one of the practices Google strongly advocates is clearly contradicting web standard recommendations and will cause your website to fail validation (we are not talking “warnings”, but actual “errors”).
If compliance with W3C standards is important to you, you will have to accept a (perhaps even considerably) lower score. If the website is supposed to score high and comply with web standards, you are between a rock and a hard place, because you cannot always have both.
The probably “funniest issue” is the use of Google fonts, the quasi–standard of using typeface today, if you reference these fonts as recommended by Google. This one–liner in your pages’ head hurts your site’s score, because it is, of course, a “render–blocking resource” (it causes your site to load slower).
If you happen to use more than one typeface (as opposed to more variants of the same typeface), Google’s recommended way of referencing these fonts even causes the validation test to howl in pain, because the “pipe”, used to set off different typefaces, is not allowed in that particular element, according to web standard recommendations. The first time I came across this issue, I couldn’t stop laughing for minutes — but now that I’m used to it, I concur that such is not funny.
Other issues may give you real headaches at first sight, though. It may well happen that you receive advice to improve your site’s performance, only to find that your site will score lower than before.
Just make yourself a cup of tea, sit back, and wait for the dust to settle. Then check whether or not you implemented the practice correctly. If you did, run the test again. If you didn’t, mend your mistake, and then run the test again.
If your score is still unbearably low, decide whether you aim for a high score (that may mean next to nothing for your ranking) or compliance with their recommendation (which may also mean next to nothing for your ranking).
It is obvious that Google is at odds with (some) W3C recommendations, less obvious (at least to me) is why so. Yet at the end of the day, it is not my (or your) business to worry about the peculiarities of the relationship between the two of them.
Conclusion
You may sit in a boat out on the sea and glimpse a light on the horizon. That’s fine, but it doesn’t mean you will also see land. If the light doesn’t change position, chances are it is a lighthouse. Then see to get there. If it keeps constantly moving, it’s probably just a cruise ship passing by at a distance. Then don’t waste your resources … and don’t lose your mind.
In other words, it is good to have tools like Google’s “page speed insights” available. Use them to try and improve your website’s performance, but don’t consider them sacrosanct scriptures that have to be followed by the letter.