Machine Learning Looks for Quality Content
- Posted on: Jun 15 2017
At Advice Media, we’ve been proponents of quality content from day one. Not only does quality content (particularly for medical practice websites) show expertise, it also aligns with the sophisticated/trustworthy image most practices need to attract new patients.
Google is a believer in quality content, as well. In 2013 the company introduced a complete algorithm update named Hummingbird that was focused largely on better understanding natural language. The next update, RankBrain, continues to try to improve that understanding.
Training line is this. Higher quality content will not only keep your site visitors engaged and likely to pass on links to your site to like-minded friends, but it will also be rewarded as Google’s machine learning continues to understand the language better. Google’s belief is that quality content better fits the searcher’s needs for information.
How Google is trying to better understand natural language
In the past, search queries could confuse Google. A word such as “best” could be misconstrued. If, for instance, the search went like this “Why are shorter skis best in wet snow?” Google could then return a link a page of the “Best short skis,” a review of ski brands. This wasn’t the searcher’s intent, but the way the word “best” is used is probably more frequent online with the “best short skis.”
RankBrain has specifically been tasked with cleaning up this kind of misinterpretation by focusing on long-tail search queries, rather than simply looking, or not looking, at specific words. Google has said it is taking more and more examples such as the “best” one above and using them to train better machine learning-derived algorithms. This is ongoing.
Here’s an example of the difficulty of the process. Google’s senior program manager, Linne Ha, who runs the Pygmalion team of linguists at the company gave this example:
“We’re coming up with rules and exceptions to training the computer,” Ha explained. Why do we say ‘the President of the United States?’ And why do we not say ‘the President of the France?’ There are all sorts of inconsistencies within our language and within every language. For humans, it seems obvious and natural, but for machines, it’s quite difficult.”
Ha’s team is focused on improving understanding of natural language. These are two factors they are focusing upon:
- What pages on the web best match the user’s intent as implied by the search query
- How comprehensive a page is in addressing the user’s needs
That second part, in particular, will reward quality content in page ranking.
How engaging is the page?
There are many different methods for measuring a user’s engagement with a page. Click-through-rate has been one of the most used, although it’s not necessarily a good judge of a page the visitor has landed upon but doesn’t click from. Google is trying to improve this area of machine learning, as well.
The ultimate judge of quality is if the page meets the needs of a significant portion of the visitors who land on it. To Google, this means these five points:
- The service/product/information they are looking for is on the page.
- They can find the information relatively easily on the page.
- Supporting information/products/services can also be found on the page.
- The page and the site build confidence that the provider is a reputable source.
- The overall page design is engaging.
At Advice Media, we strive to create and manage the highest quality websites for our practices. But we offer different levels of management/oversight, so we can’t always be on “quality content patrol.” Still, we advise all of our clients to strive for the highest quality content on every page of their sites, and on their social media feeds. If you don’t believe us, just ask Google — the search behemoth is looking for exactly the same thing.
If you have any questions about the content on your site, please contact your Advice Media representative and ask away.
Posted in: Content Marketing