I'm ripping popular landers for POPs from Adplexity. Choosing ones that have received a lot of traffic.
I'm surprised to find that their page load speed is quite slow. For example, generally around 2.5-3.5 seconds. (on GTMetrix, hosted on Amazon S3 with a CDN and GZipped)
So I'm cleaning and speeding them up by every method that I know, and getting them sometimes to speeds I'm happy with (0.3-0.5 sec).
But, there are some popular landers that rely heavily on JQuery / Bootstrap etc. With these landers im getting them down to 1-2 secs speed but hitting a wall. I'd like to get them sub 1 second, but I have a strong feeling that I would have to re-write the page in vanilla HTML/CSS/JavaScript and remove the JQuery & Bootstrap. (Which technically isn't an issue but would take time).
So my question is, should I just do the easy cleaning / optimising of the difficult landers and get them down to 1-2 secs for the testing stage, then take the winner and fully re-write it and optimise it? Is that what people generally do? Because i've read advice on other posts that you want POPs landers to be sub 1 sec but I don't fancy completely re-writing 10 landers every time I want to test a campaign!
Any advice would be appreciated! Thanks,
Gavin.
Just to make sure, do you test it several times to make sure the CDN kicked in?
Aside from that, depends how much effort it is to rewrite the landers. Basically, the more effort, the better it is to spend $50-150 test budget between them as is and rewrite the winner to improve it.
It would be strange for landers that rely on jQuery and Boostrap to be some of the better ones on AdPlexity, unless they serve a specific purpose on a traffic source like native or push where you get the click and users are more likely to wait a bit because they showed intent already.
Thanks for your reply @manu_adefy
I clean them up, remove any bullshit scripts that do nothing, add my own bullshit scripts for push/BB, offload any jquery and bootstrap to cdnjs or similar, then kraken to optimize the images. That's usually enough.
I don't test page load speed unless it seems way off.
As said above, if you find a promising offer/lander combo using your quickly cleaned lander, you can then optimize that lander once you've found it.
Speed on pops is definitely important.
It's more sensitive with this traffic type as you are interrupting people in whatever they are doing and you need to catch their attention ASAP.
The landing page will be closed faster than it has chance to load some huge external JS libraries.
Though if you need to use stuff like jQuery then use CDN, so the script is already cached in users browsers.
As for easy Vanilla JS swaps you can checkout this site: http://youmightnotneedjquery.com/, they show vanilla JS alternatives for common jQuery functions.
Hope it helps.
@jaybot
Your method is what I thought would be the most efficient. Pick off the easy wins for testing and get the load times down to <2secs. Then optimise the winner/winners. My thinking is that despite the loading speed the winning lander will show through. Then I can fully re-code the winner and get it down to 0.5sec or so.
@erikgyepes
Thanks for the link, I'm definitely saving that for later! I generally avoid using JQuery or bootstrap in my landers because of their size and how capable CSS and Vanilla JS is nowadays. (Plus if I needed a library for something there are so many great smaller more specific ones). I was surprised to see how many on Adplexity do!?
I have to keep in mind what you said about pops interrupting people, the need for fast loading, and then balance it with the most efficient way to test.
I don't find it easy to tidy someone's code on ripped landers. Then try and remove the JQuery and Bootstrap and other libraries they've used. It's far easier just to re-code the website. But I can feel myself wasting time re-coding lots of pages--slipping back into web development--rather than getting on with the task of AM.
Thanks for everyone's advice,
Gavin.
If you settle down on a specific vertical it's better to just sit down and recode it (or give it someone to do so).
Having your own landers and code is much more efficient.
Not only because you will have 100% control over everything, but also once you start scaling to multiple geos and you will be translating landers then everything will be much easier and faster.
You will be able to have just one EN version and translate it to any language.
Even better if you can create a template with the lander text in a separate file, then just load the correct file based on the language.
BTW good news about Bootstrap is that from v5 they are dropping jQuery, so things should get better with it as well.