Silktide Sitescore Review
By Brian Robson
Brian Robson, webmaster, programmer and seasoned web surfer takes
the silktide sitescore program for a test drive
Of course he had to try it out on friends and associates ...
Silktide Sitescore - what is it and what can it do?
In the Sydney Morning Herald's Next Section on 5th Nov on page 5 was a short mention of a British company called Silktide and their product called SiteScore.
Quote: "Marketing, design, accessibility and user experience are all important to a site's success. Silktide SiteScore aims to quantify some of these factors, assigning scores for each."
The purpose of this article is to evaluate the SiteScore product over a number of web sites and try to figure out how it works. The big question is this: Can SiteScore judge web sites automatically and pick winners without human assistance.
The documentation on Silktide's site "How are websites rated" explains how the product is intended to work. I tested eight web sites and my looking at the reports gave a lot of clues as to how the product actually rates web sites.
Silktide says there are five key concepts to describe a web site:
Marketing of course cannot be defined (although I'm told you can get a degree in it), but site popularity and linking to a site can be measured.
Popularity is often measured by the number of "inbound" links, i.e. other people linking to your site. This is essentially the trick that gives Google its intelligence.
Also several companies compile ratings of web sites, to work out the most popular web sites in the world. Site Score uses the ratings from Alexa. In turn, Alexa is owned by Amazon and distributes a toolbar which "calls home" with all the details of where each user has surfed to. With several million toolbars out there, Alexa compiles ratings for all the sites it can see, numbering several million. Some people would call this a gross invasion of privacy, but only the summaries are published and surfers willingly participate.
A sample of pages are analysed for key words and key phrases. Google searches are then performed on these phrases and on the name of your company or web site, as gleaned from the page titles. These tests highlight once again the importance of good titles on web pages.
- Design and Build:
The main automatic test here is putting a few pages through the validator at the W3C. I have been using this validator for several years, and it's pretty strict. Producing web pages with no errors is quite a effort, especially if your starting point is an old web site full of poor quality code and many accumulated errors. Just halving the number of errors is a realistic target.
SiteScore rates sites with CSS as superior to sites that achieve the same result with tables. It reports that pages using table-based layout should be redesigned to use CSS.
It says "effective use of stylesheets technology (CSS) allows for true separation [sic] of content and presentation."
This is indeed the current orthodoxy, but the whole idea of HTML was surely that the markup tags were intermixed with the text. But, unlike a printed book, the user now controls the layout with the browser settings. Unfortunately CSS has been abused by web designers especially by using tiny fonts that are too small for normal people. The whole concept of CSS is suspect when accessibility guidelines say that web pages should still look OK when the user has disabled CSS or when the CSS file has gone missing.
SiteScore recommends frames be removed as these are considered old design. This is a great idea, and SiteScore summarises the problems with frames in a nutshell - difficult for search engines, hard to bookmark and frames compromise usability/accessibility for users with disabilities. I have never used frames, but it's been a struggle to get others to take their frames off, as so many web design tools lead you to believe that complexities such as frames are quite normal.
In the United Kingdom, a new disability act came into force on 1st October 2004. Most English web sites are now illegal under this act, and I'm sure the government will not know where to start their enforcement. There is nothing like passing laws based on the post-modernist assumption that technology can fix everything.
Basically, without actually learning anything about accessibility, the more complex a web site is, the less accessible it will be. It's clear in SiteScore that the Alt tag on all images is very important.
It's a pity the UK government didn't legislate for usability instead of accessibility. That might have fixed a few web sites that are best described as crimes against humanity.
- A Satisfying User Experience:
The SiteScore definition praises sites that provide "useful services or valuable content".
Factors here that can be tested for include User Forums, a Log-In area, and a Downloads area.
Good features said to be rewarded by SiteScore include the menu items About Us, Contact Us and Search.
Involving the users with interactivity is highly rated and in essence means the use of forms on your web pages. If you don't have a form for users to fill in, you score badly.
SiteScore rates the speed of a web site as a major part of the user experience and comments often about fast servers. SiteScore does not say it, but if you have big HTML files and giant graphics and only a dial-up connection to the net, getting a faster server will not help one bit.
- User Ratings:
As explained on Silktide's web site, this is the only non-automatic test. Instead, real net surfers are asked to look at and rate websites (picked at random) between 1 (very poor) and 10 (excellent). However some rules apply. No web site rated less than 4 is shown to users, and the choice of sites shown to users is deliberately biased to showing those already rating over 8.5.
Let's now examine the table of results. All scores are numbers where 1 is
terrible and 10 is excellent. It appears that ratings less than 4 rarely
occur, except where sites have been automatically downgraded by their adult
|Design and Build||8.0||7.0||7.4||8.2||7.9||8.9||9.3||9.3|
The following is a list of the sites that were examined:
This is the Non Smokers Movement of Australia at www.nsma.org.au The scores are typical for a small, vanilla website. It has the Search feature as provided by Google, and works ok. The test results do not reveal that other sites, especially blogs, link deep into this site to pick up pictures of celebrity smokers.
This is the Bondi Beach Home Page at www.bondivillage.com - this is by far the largest and busiest of sites about Bondi and was professionally designed in 1997. Since 1999 it has been expanded mainly with picture galleries, and the standard of the code and navigation has been improved, but very little text content has been added. In fact, under guidelines about writing for the web, the quantity of text has generally decreased.
It is very widely linked and usually comes up first in search engines, looks terrific and is reasonably lean and fast and yet it scores poorly as a satisfying experience.
In fact, earlier figures were even worse, as SilkScore managed to test the pages having the most errors; and the site was slower on the day it was tested.
So, I bit the bullet and fixed most of the errors found by the W3C testing. A re-test with SiteScore, and errors dropped from 72 on 3 pages to 10 on 2 pages. Also I fixed most of the missing Alt tags which seems to be a determiner of the accessibility rating.
With these tasks completed, the Design score increased from 6.6 to 7.0 and the Accessibility score from 7.1 to 7.8.
However, the number of links claimed to be in Google decreased, and the overall popularity dropped from 496,000 in the world to 543,000 with the end result that the Marketing score decreased from 8.7 to 8.4.
This woman is a world expert on Apostrophe Errors and an Aussie country music fan as well at www.sharoncolon.com.
Sharon's site is hosted in Europe and was the fastest of all sites tested, but not surprising as SiteScore is testing from England.
Some separate testing of Sharon's country music pages said 4 pages had virtually no text and that text should be added to these pages. In fact these pages are just thumbnails where the users click on each thumbnail to see the bigger picture. No text is required, the users don't read it anyway. Clearly the concept of judging the balance between text and graphics is not easy.
Eventually I discovered a site linked from Silktide which offered the opinion that no web page should have less than 200 words of text.
This is Casey and Lowe, a small firm of Archaeologists in Sydney. Their web site www.caseyandlowe.com.au has only just been completed. It is just starting to appear in search engines and is not linked from anywhere yet.
Consequently, it has rated only 3.3 on the Marketing scale, and thus rates poorly overall. Clearly things will improve. The site itself has a good balance of pictures and text, a consistent interface and colour scheme across all pages, and a lot of PDF reports that users can download.
This site is www.danbyrnes.com.au and it's about all things ancient and esoteric with an emphasis on genealogy, timelines and the history of convict ship owners and slavery. The site has good content but the pages are too long and the graphics are often oversized. Not much relevant information is presented "above the fold". Navigation is frequently inconsistent or missing, and the site relies heavily on search engines such as Google to index deep into the site and then refer users to the correct page, according to their search keywords.
The site does use CSS and thus rates well for Design. It uses a form to collect email contacts and thus rates well for user satisfaction. In reality, it is hard to believe that this site rates higher than the Bondi Beach Home Page for user satisfaction, but it does.
The Heritage Office
Just for comparison, here is a government web site, properly resourced and professionally programmed. It's surprisingly busy, with loyal users, and is getting faster as unnecessary graphics and features are gradually trimmed.
It's at www.heritage.nsw.gov.au and contains thousands of pictures of heritage in NSW.
This is a small lobby group called the Doctors Reform Society. I cleaned up their web site at www.drs.org.au a couple of years ago but there have been a lot of changes since then. The site looks bright and cheery and is updated regularly.
Overall, it scored the best of all the sites I tested.
This is the Tamworth Rage Page at www.tamworthragepage.com and it's the biggest country music fan site in Australia. It's built entirely by one person using the dreaded FrontPage but it's very popular. Despite literally hundreds of errors in the actual HTML code, SiteScore rates it 9.3 for Design which is truly amazing. As the site grows, the navigation gets more and more convoluted, but the fans are loyal - it's the best source of country music news in NSW.
As you can see, the SiteScore tool is patchy and by no means mature, but it's a step in the right direction. SiteScore claims to have tested 10392 sites, and only 750 have rated 7.5 or better. None of the sites I tested have reached 7.5.
My suggestion would be to get rid of the User Rating section entirely, and concentrate some research on deciding if a site looks any good and has meaningful content and navigation "above the fold".
At present, SiteScore is not sure if the same navigation and images page after page is planned consistency or just plain boring.
Using SiteScore has made a few overlooked things clear to me. For example if you have a page with text that reads "Sharon Colon proudly presents My Holiday Pictures" it will get indexed under all these words, even quite unintentionally under the phrase "proudly presents" which is just a throwaway.
I have discovered that a Company's address, suburb or motto, repeated on every page becomes a powerful thing in indexing, again quite unintended.
Readers may care to test the tool against their favourite sites, and especially some with splash pages and lots of flash.
Also, readers should expect the tool to improve over time, as the rating techniques are honed. Testing across several sites will reveal the size of this product - there is quite a range of diagnostic messages and hints.
It is easy to be dismissive when your own site gets a poor rating, but how come all those top sites get such high ratings. The starting points to improve your own site are the quality of the code (including adding those pesky Alt tags to all images) and checking the indexing of your site in Google for both the correct title and the expected keywords.