The major search engine companies have been struggling for some time to develop a quality index algorithm (QIA) which can better determine which sites a searcher may be interested in pursuing. Google’s most recent deployed solution was the link strategy. Of course as soon as people became aware of the fact that quality incoming links were important several products quickly emerged:
Reciprocal Linking - this is the ‘I link to you - you link to me’ strategy. Reciprocal links were quickly spotted by the search engines.
Link Farms - in an attempt to deceive the search engines, link farms were set up where A linked to B, B linked to C, C linked to D and D linked back to A. A bit tougher to discover in low link numbers, but if a website did en masse link farming they still got noticed!
Pay for Link Programs - If links are so important, why not pay for them? These programs started generating revenue almost immediately, and were quickly discouraged by Google and others.
Recently in the news is the widespread “punishment” of companies that accepted pay for link programs but failed to use the nofollow attribute or some other means to delineate those links as links for people only, not search engines. Aside from the widespread complaints that it was bound to stir up, many people got worried that their sites might be harmed because a company they advertised with did not use the nofollow attribute or other method to prevent search engines from counting their links.
Google was pretty quick to come out and say that this would be unlikely. They would go after the paid link companies that they felt were violating Google’s policies, not punish unknowing subscribers to such links. Even though it will be awhile before the dust settles, Data Choices chooses to predict the next step in search engine results pages (SERPs); the Quality Index Algorithm.
QIA
The Quality Index Algorithm will develop after the end results of the link program do not substantially identify the highest quality results relative to the searchers’ request. Some people have suggested that it will be a time on site indicator. Of course such a simple indicator could easily be manipulated by parking numerous machines on a single website. Can’t you already hear all the companies advertising? We, at Park-a-Lot will put 250 computers on your website and leave them there all day - for a fee! No, a simple time on site algorithm will not suffice. Instead, time on site will need to be mixed with a depth of visit indicator as well. Suppose someone comes to site A and spends 21 minutes on that site, visiting three pages. Then he goes to site B and spends the same amount of time there, but visits seven pages. Which site did he like more? The first, because he spent so much time on a few pages, or the latter because his depth of visit was more?
Quality Index Algorithm
In the above graph we see length of stay along the X-axis. Depth of visit runs along the Z axis from the far wall towards the viewer (10 pages is closest to you, one page furthest away). The QIA runs along the Y axis. The QIA is virtually zero at only one page depth. Likewise it is close to zero at less than a minute per visit. Generally as length of stay increases and depth of visit increases (comes closer to the viewer) the QIA goes up. However, as discussed above, there are various “wrinkles” in the algorithm, which will be proprietary choices made by the search engines depending upon how they balance their algorithms.
This is where the search engines struggle. It will be pretty easy to set a ‘low quality’ setting for sites that get only one page view, or those where the length of stay is less than half a minute. But as the depth of visit increases along with the time on site, the engines will actually need to analyze the type of content and the layout of the content - something they are not currently prepared to do!
Type of content comes in many forms, but let’s consider flash versus text. A lengthy, high quality text article may keep viewers on site for quite some time. But an incredibly popular one minute flash video is likely to keep viewers on site for - well, about a minute. But that doesn’t mean that the flash video is any less useful than a page chock full of great information! So how does a QIA handle this?
If we know that the average viewer goes to this specific one minute flash site and stays the full minute, we can reasonably presume they sport a good quality flash. If most people go there and stay five to ten seconds, we know the video was not compelling enough to keep people watching.
It is pretty much the same for a text site. If the average viewer spends fifteen minutes on a similar (or “control”) text site and they spend that much time on the site they just visited, quality must be relatively high. And if they leave in less than a minute then they were not compelled to stay long enough. Although this is a very simplistic view of a complex situation, it really breaks down into the following:
Time on site will be measured against depth of visit. Different search engines will use different algorithms, but the most successful ones will be the ones that balance time on page, while still getting depth of visit.
Sites will need to be categorized by type of content, so apples are compared to apples, and oranges to oranges
Site traffic will be an important factor as averages of the aggregate will become more important
Individual site averages themselves will be important! If most viewers spent two minutes on a certain page of your site last month, but now are spending five minutes or more after the content was retooled, the QIA will rightly presume that the quality of the page has increased
Certain drop off figures will be used in the QIA. For instance, page views of less than six seconds will likely not be used in determining time on site for quality averaging purposes. Additionally if most people visit only one or two pages, that will be weighed much less than the people who visit three or more pages. In other words there will be a minimum set for depth of visit in determining averages.
Finally, and perhaps most importantly, the search engine bots themselves will be retooled to estimate the average length of time it would take to “consume” a given page, whether that is reading text, watching video or a combination of both. This consumption number will play a vital role in that actual live consumption from searchers will be compared to these figures.
At first the QIA will only weigh in at ten to twenty percent of the SERP ranking process. But, as the algorithms improve and the results truly do begin to represent quality, that weight will increase dramatically. No; links, traffic and other current methods aren’t going away. But once QIA is alive and working we will see a dramatic decrease in their weighted value as the weighted value of the QIA increases!
Ultimately what this means to the website owner is that compelling, quality content will become even more vital to your companies’ success in the future! This report was first published on 10/27/2007
http://www.datachoices.com/public/QIA.html
http://www.datachoices.com/public/QIA.html
Data Choices
We compile. You decide.
http://www.DataChoices.com
Keyword Tags: algorithm, Google, html, linking, Search Engine Marketing, SERP, video
No Responses to “QIA - The Quality Index Algorithm - Coming Soon to a Search Engine Near You”
Please Wait
Leave a Reply