mobile pagespeed 38

COPPERRAVEN

The client's shop scores 38 on mobile in PageSpeed Insights and 71 on desktop. It is an e-bike battery shop, around 120 products, built with Elementor Pro on Hostinger shared, LiteSpeed Cache on with the page cache and the image optimisation turned on. I think I build it from one of the Elementor shop templates like I do since years and the client was happy untill he ran the test himself.

Now he says the builder is the problem. He read somewhere that Elementor is slow and he wants me to rebuild the whole thing with the block editor, wich is two weeks of work he does not want to pay for.

I say it is the host. The TTFB in the report is 1.9 seconds on the product pages and 1.4 on the home page, and I don't see how the builder can make the server wait two seconds before it sends anything. The LCP is 6.8 seconds, total blocking time is 890 ms.

But I don't know how to prove it to him. My other shops on the same template score between 45 and 60 and they are on different hosts, so this doesn't tell me much either.

So what do you think with these numbers, builder or host? And if your host is also shared, what do you score on a product page with Elementor? I would like to show him the real numbers from other shops and not the opinion of one blog post.

christopher

Welcome and there was a builder thread here earlier this week you should read, same argument.

I'd say host plus images. Two seconds TTFB isn't Elementor.

My plumbing client had almost the same numbers on the old host and the builder never changed.

We moved him and the mobile score went from 41 to 68 before I touched a single image.

troygray88

christopher wrote
I'd say host plus images.

No, it's the builder. Elementor on shared hosting is always slow, every one I've seen scores in the 30s and 40s on mobile, and you could move your host three times and it should still land there.

evan

Neither, until you measure it properly. Why is everyone in here guessing from a score?

the report has a line under diagnostics, reduce initial server response time and that's the bit that tells you. Run the test twice on the same product page so the cache is warm and if that line still says around a second or more its the server, but if it drops to 200 ms and the LCP stays at 6.8 then it's the front end and the host cant help you. I can read 1.9 in his post, but one cold run on shared hosting should not count.

COPPERRAVEN

I run it three times warm. Server response stays on 1.9 s, but the LCP element is the hero banner and this is a PNG of 2 MB so it is both not "always the builder". I can post the screenshot of the diagnostics if someone wants it.

vandana

Slightly beside the point, but the 38 itself is what your client is fighting about not the shop.

I'm seeing this with every client since years. They run the test once on the phone. See an orange number and panic.

What I am doing now is I scroll past the score and show them the box on top, the one with real user data from Chrome. If that box says the Core Web Vitals pass, they calm down because that's the data from their own customers and the lab number is a simulated slow phone only.

If the shop has too little traffic the box is empty then you have to fall back to the lab numbers and have the discussion anyway.

Did your client read anything beyond the one blog post?

ivanvallejo

Careful with "they calm down" as the goal. That box is Core Web Vitals and for a shop thats the one thing Google uses so you'd have to read it as the result and not as a sedative. A 38 with green field data is fine, I'd leave it alone. A 38 with a red LCP in the field data is a real problem no matter how the client feels about it because that's the phones of the people trying to buy a battery.

For what it's worth, I'm on Hostinger too with the coffee shop, business plan, Redis on, and the server response just sits around 400 ms, so 1.9 isn't normal even for them. If his shared plan does that warm I'm pretty sure a plan up fixes half of it before anyone touches the builder.

vandana

That is what I meant, I was only putting it badly.

One thing for your setup... I am on Hostinger shared also with the t-shirt shop and the warm server response is around 700 ms so 1.9 s on your end would make me check whether the LiteSpeed page cache is hitting at all. You could look at the response headers in the browser for that.

troygray88

Fair enough, the 1.9 is the host, this is the bit I got wrong. I went back to my own Elementor sites and the slow ones all have big images too, so maybe I blamed the builder for that.

evan

Yeah, I've seen that pattern enough. Both, yes, but start with the host because no builder setting on earth takes 1.9 s off a server response and the 2 MB banner is ten minutes of work.

Did vandana's header check show a cache hit? That's the bit I'd want to know.