The philosophy that guides Bing algorithm revolves around throwing up most useful results right at top of the relevant Search Engine Results Page. To put it in simple terms, the engine does so by ranking the results in apparent descending order of importance. This is based on the clicks any given result gets. Of course, specific answers, in turn, need to be competitive and comprehensive with the results they displace so as to justify their position within that results page.
This may seem rather straightforward. Occasionally, there are instances wherein answers don’t exactly match user intent despite receiving a fair amount of clicks that justify their higher position. Bing deems such results defective in nature. To tackle them, it has created dedicated models that will deliver only the most pertinent and relevant results. Defects might still make their way through so the engine employs certain models for minimizing unhelpful and irrelevant answers from surfacing in the search results.
For example, most people for the common query (elephant), will opt to interact with the following elements on SERP:
- An images answer, which showcases evocative images
- Wikipedia result
- A video answer that highlights popular videos
Deep within the internal indexes of Bing, there’s another answer looking to vie for user attention. An answer has been identified by the product index for (elephant) posters. Even while one cannot buy elephants, Bing can well identify posters you could purchase along with this particular answer:
Its defect classifiers will mark it as a palpable defect since it doesn’t go with the user’s outward intent. It will get blocked from that page though a large percentage of users may well click on it. This goes to highlight a peculiar problem area of relying on higher click rate solely. In the above case, people might be clicking owing to the graphic though they lack any interest in buying a poster.
The defect classifier of Bing will use various other signals apart from click rate (these include the way people have generally engaged or dealt with a query category) to determine whether an answer is relevant or not in the context of other results and then block the same from the results page.
Another set of queries susceptible in terms of ambiguous intent as well as defective answers happens to be navigational queries. These queries include users typically navigating to any single website or a webpage.
In the simplest of cases, there is possibility of intent mismatch when the Bing federated indexes like its video index carry content, which matches that query. The results seem relevant in isolation on basis of simple ranking criteria. However, the user has no real intent for the specific content. The answers generated are obviously clear defects.
In a federated engine, an array of indexes and query answers run in sync (in parallel) so as to produce results in a matter of milliseconds that strongly fulfill the intent. The Bing defect classifier can use data on how the Web, news, market and finance answer will interpret this particular query for deciding if the local answer happens to be defective ((least competitive with the content on that page). In such case the ranking as well as the Bing defect model will agree to block it.
Other source of glaring defects can be attributed to poor quality (search) results and not just misinterpreting the users’ intent. This happens in cash an index gets rather lax at letting a partial match between the user query and its actual content. At times, ambiguous queries are difficult to handle.
Bing constantly works to enhance its defect classifiers so as to improve the search result relevancy. SEO experts must keep track of the evolving Bing algorithm methodology to target higher position in SERP for their pages.