We know that Google and Bing look at hundreds of different signals, such as the number of backlinks, to decide what pages should show up at the top of search results. But have you ever wondered how they weigh those signals to get the very best search results possible?
Believe it or not, much of the information that they use to decide the weighting is from human testers! Recently, a copy of the guidelines that Bing’s testing staff use to judge webpages was obtained by SearchEngineLand.com. These guidelines reveal exactly how a page is judged.
Is Your Landing Page Good Enough?
Most of the time, it is impossible to boost your website to the top search result spot just by increasing the number of link pointing to your site — you need to meet all the criteria for a good landing page. Here is what you have to consider:
Bing first considers the searcher’s intent: it could be:
1) Intending to get information
2) Intending to navigate to a specific website
3) Intending to make a transaction
Even this information is incredibly useful. To ‘hedge their bets’, Bing would want to diversify their search results and show results that satisfy each of these intents. Maybe it’s not super important to focus on all three of these on your landing page, but instead to focus on just being, for example, an authoritative source of information.
Range And Depth
Bing then decides whether the landing page has enough of a range of information, and whether that information is in depth enough. This means it will be difficult for your page to rank #1 if you only have light content and a tiny website.
Bing also wants to know if the landing page is appropriately trustworthy, or ‘trustworthy enough’ for the user. This is why new websites are harder to rank than older ones, bigger websites are worth more than little ones, etc.
Finally, the overall quality and appearance of the landing page is judged. Obviously, your page has to be well designed and easy to navigate. There is even a special flag that testers can mark your website with that tells Bing: “Don’t rank this site, it’s too inaccessible!”.
This isn’t one of the main criteria, but Bing testers also mark landing pages for how important it is that they should be fresh. In other words, a page about who won the 2004 Olympics will be marked “freshness not important”, whereas a page about who won the 2012 Olympics will be deemed “freshness very important”.
When a website is judged, it gets an overall score from “1. Perfect”, to “5. Bad”. Perfect pages are pages considered the definitive or official webpage, and must perfectly address the intent of 50% of searchers for that query. Bad pages address the intent of less than 1% of searchers, and are either domains not being used, domains that use other website’s content, and domains that use spam techniques or install malware. There is also a special category “detrimental” for certain types of pages such as those with adult only content.
Machines Learn From Humans
Bing testers look at as many pages as they can, but of course it’s only the smallest fraction of the search results that they could possibly judge So how is this information useful? Like I said, Bing (and Google) probably use this information to help decide which search signals should be considered the most important. A popular approach to machine learning is to emulate human intelligence by taking into account all relevant features (search signals), then learn and adjust based on human feedback. This means it is super important to not write your website for search engines, since probably no one really knows exactly how they are weighted, and instead to write your website for humans.
If you’re interested, SearchEngineLand also wrote a piece about Google’s quality testing.