Today the loading speed of a website is essential for it’s success.

Google, Yahoo and Amazon made tests that has shown the importance of performance.

In short:

- Google slowed down their page for 500ms – 20% drop in traffic

- Yahoo slowed down their page for 400ms – 5-9% drop in full-page traffic (users leaving before site finished loading)

- Amazon slowed down their page for 100ms – 1% drop in sales

20% of traffic lost for Google … immense drop.

How To Optimize The Front-End To Get The Maximum Loading Speed

ATTENTION: this post is really very technical. If you want to boost your websites loading speed but you have no clue about what I am talking about, feel free to contact me. I am sure I can help you.

Front-End Optimizing is optimizing a website with different techniques to achieve the fastest possible loading speed. In this case only the Front-End counts.

The Front-End is the part of the website that is transferred and executed/rendered to the client (browser) and contains at least the following four parts:

- HTML code
- CSS code
- JavaScript code
- Image files
(optional media files like .swf, .mov, .mp3,…)

How does Front-End Performance Optimizing Work?

The different parts of the optimizing process are:

1. Common HTML Programming Tasks

- Avoid HTML errors (W3C Validation)
- Delete comments from HTML document

2. Common CSS Programming Tasks

- Avoid CSS expressions

3. Optimizing HTML

- Optimize the order of styles and scripts
- Avoid usage of tables for website design
- Avoid usage of inline styles
- Avoid usage of inline JavaScript

4. Optimizing CSS

- Remove unused CSS

5. Optimizing JavaScript

- Remove unused JavaScript

6. Minimize HTTP Requests

- Combine external CSS
- Combine external JavaScript
- Use CSS Sprites to reduce the amount of images to download

7. Compressing CSS/JavaScript/image files

- Minify combined CSS
- Minify combined JavaScript
- Optimize image file size with tools like Yahoo Smu****

8. Reduce download size and increment number of concurrent downloads

- Serve static content (CSS/JavaScript/image files) from a cookieless domain
- Parallelize downloads across hostnames

9. Server-Side Compression and Browser Caching

- Enable compression (mod_gzip/mod_deflate on Apache web server)
- Leverage browser caching (mod_expires on Apache web server)
- Specify a Vary: Accept-Encoding header (mod_headers on Apache web server)

These are the essential parts to get a really fast loading website.

The following online tools can help you to get the best results:

Combining and Minifying CSS/JavaScript files – http://scriptalizer.com/

Minifying JavaScript/CSS files – http://compressor.ebiene.de/

Compress Graphic files – Yahoo Smush It (Google for it)

Minify CSS/JavaScript and Compress Graphic files with Google Page Speed for Firebug (Firefox addon)

Another tool to check the loading time:

Compare loading speeds of websites – http://faster.1click.at

If you need some informations about cookieless domain then visit http://www.cookielessdomain.com/blog/

And the authoratitive source for these statistics is??

Yahoo did this tests.

You can check here:

http://www.slideshare.net/stubbornella/after-yslow-a

or here, if you want:

http://www.slideshare.net/stoyan/yslow-20-presentation

Ok

I must be missing something, such as some actual test results?

There is no actual since the statistics would be nearly the same.

Bye

So the "statistics" are merely guesswork then.

The stats about Google, Yahoo and Amazon are not guesswork.

Those were done by Yahoo.

You just need to use Google Page Speed addon and YSlow to find out more about what I am talking about.

The Developer Network of Yahoo is a good source, too.

And recently Google released their mod_pagespeed for Apache. This was not done for fun.

The stats about Google, Yahoo and Amazon are not guesswork.

Those were done by Yahoo.

Sure, but where's the proof?

The whole world of Search is full of unsubstantiated "facts", guesswork and fantasy from "experts", some of which rival Tolkein in their imaginative outpourings.

Without some actual information to backup the claims, it's just another "8 out of 10 cats prefer" scenario

Did you see those Slideshares?

Maybe you ask Yahoo and Google to get a piece of paper with the proof.

Ask Stoyan Stefanov:

http://www.slideshare.net/stoyan

One more for you:

http://www.slideshare.net/stoyan/wpo-pubcon-2010

Google search : http://www.google.com/search?sourcei...0ms+20%25+drop

Did you see those Slideshares?

Yep and I'm stil no wiser in what or how they came to the conclusions.

Are the "results" for:
The "home" page?
Specific pages?
A range of pages?

How many users?
What browsers?
etc etc.

It just another lot of vague, ambiguous "facts" about "page speed" that is designed to confuse and convince users that they MUST use these "tools" to improve whatever it is they need to improve.

The search world doesn't any more crap like that, it simply needs a very large injection on commonsense!!!

Yes, Mr. Super Clever.

It does not matter that Yahoo Engineers mentioned it, in your eyes it is crap.

Ok, then it is crap, because YOU (Mr. Super Clever) said so.

Ask Yahoo,

thanks, bye.

NOONE ever said you MUST use this tools.

Oh and btw,

if YOU think this thread does not have any useful informations then DELETE it.

@chrishirst

Check out this video:

http://video.yahoo.com/watch/4156174/11192533

There is your proof, no?

Originally Posted by CSM

@chrishirst

Check out this video:

http://video.yahoo.com/watch/4156174/11192533

There is your proof, no?

The thing about slide shows is that they tend to only outline a presentation; a lot of the details and background information are missing. The video fills in some of the gaps, but there is still one piece of information missing: the data.

If you say something like:

Amazon slowed down their page for 100ms – 1% drop in sales

How do I know that 1% isn't within the margin of error? What's the sample size? How was the test conducted? How do I know other factors didn't influence the results?

Oh and btw,

if YOU think this thread does not have any useful informations then DELETE it.

Isn't pointing out that information is useless useful information in and of itself? (I'm not saying that is the case in this thread)

Deleting threads that you think are useless (not including blatant spam and total nonsense) isn't really in the spirit of a forum. Explaining that some information maybe should not be taken at face value is.

Regarding the original post:
99% of what I do involves backend coding, so maybe there is some subtlety I'm missing. How does using tables impact loading time? I haven't seen anything to suggest that using tables increases the number of bytes in a page or the amount of time it takes the rendering engine to process the page.

Nope

Still no verifying data just Yahoo! people giving opinion.

Now had you posted this which also points out that Amazon did A/B split testing to reach their conclusions it helps to make the claims somewhat more quantifiable.

Here of course we are dealing with perception and pages that people already know to be "quick", so adding a percetible delay can certainly create some impatience with users.
I do it myself, if a page I expect to load in a certain time frame doesn't, I will close the browser window and try again, A/B testing cannot detect that, it will only determine that the slower page was closed, extrapolating that to actual loss of revenue is going to be somewhat inaccurate, because while people are often impatient, they are also persistent, and provided what they are wanting is on that page they will try again.

Now, had they "slowed" a page and run a comparative sales analysis that actually determined a drop, the "revenue loss" suggestion would be substantiated. Without that, it is supposition that the page closure actually led to a loss of sales.

There is less patience when shopping "virtually" as opposed to the high street because in high street store there is always something to look at, so provided your pages give the viewer something to look at or read relevant to their query they very likely will wait that extra half second or so.

If you want to find what YOUR customers will tolerate rather than what the Amazon customers expect you should do your own split testing.

Uhm well, since I do not sell anything online a split test does not help me much.

The thing is just this: the sites have to be fast, if not users will go somewhere else.

Faster sites, more sales (in general).

The thing is just this: the sites have to be fast, if not users will go somewhere else.

And do you?

My sites are optimized.

If I visit a site and it's not loading that quick it can happen that I am leaving