Semantic content: images with alt attributes vs plain text? The CONCLUSION

For any of you who have been tracking our quest to see if semantic markup truly does have an effect on natural search engine rankings we can confirm we now have a pretty concluisve set of results.

We ran three different versions of content: (H1/P are the tags used to mark headers)

1. "Frank"
   (H1: image with alt and title)
2. "Fred"
   (H1: as plain text)
3. "Roger"
   (P: as plain text)

The analysis...
Roger was the clear loser on Google and failed to get indexed after initially ranking quite high (second to Fred for much of the experiment).

Roger was however the first to be indexed by Google, Yahoo AND Altavista. So if it's quick wins you want Roger is your man.

Frank ran a clear second to Fred for much of the experiment (not unsurprisingly and thankfully the suspected result).... HOWEVER... Over time it has proved pretty fairly conclusive that Frank and Fred had NO significant difference in ranking on Google.

In fact 2 of the 5 search engines completely ignored all of our pages so we cannot be entirely conclusive. Ask and Windows Live did not index ANY of the pages at ANY point throughout the 3 month experiment so clearly their algorithms rely more heavily on inlinking or natural spidering - of which none formed a part of our experiment.

The conclusions?
Well to be honest... Pretty unconvincing. It has to be said that at the time of writing (and when we closed the experiment) Frank (H1: image with alt and title) was the ONLY page naturally listed in Google's results.

Roger(P: as plain text) listed consistently best on Yahoo.

Frank & Fred had the most consistent results overall suggesting that, in our opinion, semantic content DOES make a difference. Although we are truly shocked that it didn't appear to make as much difference as we suspected, especially considering accessibility laws and the use of semantic tags to help screen reading browsers to navigate.

Our next experiment will be based around meta information in a quest to see how search engines actually use meta information and it's significance for companies trying to achieve high natural rankings in the major search engines.

So in our first experiment to uncover SEO secrets with "No bullsh*t just hard & fast evidence of what works best" proved pretty inconclusive... Sorry guys! :-)

Caveats to consider...

1. Each page was given similar content (same volume of keywords completely irrelevant to what the test website was about and each other).
2. Each page had no in-links from other sources except those that appeared in the Code Required blog (however we cannot account for links to the pages from other blogs etc - at the time of writing we were unaware of any and the pages have been removed from the test site)

Semantic content: images with alt attributes vs plain text? Part 2

OK so interesting results from our little test case... Although it looks like Google is certainly acting as expected for natural rankings with the way we markup content we've had some interesting results from the other guys at Yahoo & Altavista where it looks as though our mate Roger (un-sematic plain text content) is certainly leading the way!!

Page name	Position
	Google	Yahoo	Altavista	Ask	Live/MSN
Frank (H1: image with alt and title)	3	n/a	n/a	n/a	n/a
Fred (H1: as plain text)	1	n/a	n/a	n/a	n/a
Roger (P: as plain text)	2	2	2	n/a	n/a

I must say - I'm a bit concerned by these results as it certainly doesn't bode well for the way we've been told we "should" code so often!

Watch this space for further updates!! Feel free to drop Jon an email if you want to see more comparisons.

You can see the pages at:
http://www.mycardioworld.com/tests/frank.html
(header as image with alt and title attributes)

http://www.mycardioworld.com/tests/fred.html
(everything as plain text)

and...
http://www.mycardioworld.com/tests/roger.html
("un-semantic" plain text)

Semantic content: images with alt attributes vs plain text?

One of the big arguments I constantly have about SEO is whether semantic content as images (with appropriate alt text) or simple text values will rank differently.

For example - which is better for SEO?

<h1>header</h1>

Or...

<h1><img src="header.gif" alt="header" /></h1>

...And does the H1 tag actually make a difference?

<p>header</p>

Not surprisingly it's an incredibly difficult subject to find any solution for so here at Code Required we're currently running a simple test to see what happens...

You can see the pages at:
http://www.mycardioworld.com/tests/frank.html
(header as image with alt and title attributes)

http://www.mycardioworld.com/tests/fred.html
(everything as plain text)

and...
http://www.mycardioworld.com/tests/roger.html
("un-semantic" plain text)

We'll be monitoring the situation over the coming weeks on all the major search engines and posting updates here - in the meantime if you would like us to add other versions or have any comments please drop Jon an email.(For example should we do a sIfr version? should we do a version without title attributes on the images? etc).

Page name	Position
	Google	Yahoo	Altavista	Ask	Live/MSN
Frank (H1: image with alt and title)	1	n/a	n/a	n/a	n/a
Fred (H1: as plain text)	2	n/a	n/a	n/a	n/a
Roger (P: as plain text)	n/a	n/a	n/a	n/a	n/a