During the first week of November 2019, Google held its first Webmaster Conference at the GooglePlex in Mountain View, CA. The conference was designed to help webmasters improve their site’s content by optimizing it for search. Several topics were discussed, including synonyms for search, duplicate content, and the announcement of a new language processor: BERT. Google relies on all three of these tactics for information retrieval to properly rank websites. Knowing how to utilize these tactics can help your sites rank higher in the search query! Keep reading to find out how.
Synonyms for Search
Information retrieval is about matching the keywords in a user’s search to the content that the user is looking for. It should be noted that keywords may be in the title, body, or links and can also be measured by how frequently they occur. When a webmaster is optimizing her site for search, she makes a list of synonyms that a user could possibly look up on a search engine that may lead a user to her website. During the Synonyms for Search lecture, the speaker discussed how language varies in meaning and context. For example, the letters “GM” could reference General Motors, genetically modified, or general manager depending on the context around it. The speaker also gave an example of how using “or” in a synonyms list is beneficial. When making a list of potential synonyms, the webmaster wants to be as thorough as possible. The following example used in the conference gives a good idea of how many synonyms the webmaster can use:
“Cycling tours in Italy” → “cycling OR cycle OR bicycle OR bike OR biking OR tour OR holiday OR vacation in Italy OR Italian” (Seer Interactive, Clark).
Jackie Chu wrote an in depth review of the conference that highlighted a lot of important points. She notes that Webmasters must also take into account ‘sibling’ words. These are words that serve the same purpose, but do not carry the same meaning, such as cat/dog and contact/address. Non-compositional compounds can also interfere with user searches. For example, if a user was to Google “York hotels”, they would not be met with hotels in New York, but they could search that phrase for Jersey or Vegas. Some words have to be used in conjunction with one another to be contextually sound and bring up what the user is actually searching for. New this year, Google users will be able to search for emojis. Google receives roughly one million emoji inquires a day. Google spent the past year optimizing its language analyzers to understand searches that included “*emoji* meaning” to better meet the needs of users.
Here at Aronson Advertising, our webmaster team comes across duplicate content all the time. Duplicate content is an issue because Google does not know which page would be better to include in search results or rank for query results. Directing link metrics (trust, authority, anchor text, etc.) to one page or keeping them separate between two or more versions can also get complicated.. All of these factors can lead to traffic loss and low rankings. According to the SEO experts at Moz, traffic loss comes from the fact that search engines rarely show multiple versions of the same page so that the user can get a plethora of options. If the search engine can not decide between versions of a website’s content to show, it will have to guess which is the better option, which dilutes traffic to both pages. At Aronson, the solution we use most often when we come across duplicate content is 301 redirects. That way, the “duplicate” page links to the original content and they stop competing with each other. This will greatly improve the original page’s ability to rank higher on search results pages.
Finally, Google unveiled its latest language analysis project, BERT, or Bidirectional Encoder Representations from Transformers. BERT is special because instead of only reading from left to right, the software can also read right to left, giving it a “deeper understanding of language context and flow than its single-direction counterpart” and allowing “the model to learn the context of a word based on all of its surroundings”, according to Rani Horev in his article for Towards Data Science. BERT uses two separate processes, “an encoder that reads the text input and a decoder that produces a prediction for the task”. To show off what BERT can do, the speaker Googled the phrase, “Can you get medicine for someone else pharmacy”. Before BERT, the top result was an article from MedicinePlus about how to, very generally, get a prescription from your doctor filled at a pharmacy. With BERT, the top result was a HIPPA forum that answers the question, “Can a patient have a friend or family member pick up a prescription”, which is most likely what the searcher was looking for. BERT’s ability to read the context of a search is unmatched by its predecessors.
Those who attended the conference were able to learn a lot and share the information and key takeaways with their readers. Zaine Clark had helpful ending remarks for his readers at Seer Interactive. The first and most important one being to focus on people, not search engines. Webmasters must be able to think about the context behind users’ searches and include all possible synonyms to have their content rank above its competition. Secondly, the speakers urged webmasters to keep their site clean of duplicate content so that Google can best deliver their content. Finally, the best way to make sure your site is serving its purpose is to not chase the algorithm. Google is constantly evolving the way it delivers content, so as long as your website is technically sound and you’re posting content that fits your audience, the content will flow to your users.
We are always looking for fresh talent here at Aronson Advertising! Browse our current openings to see how you can kickstart your career. If you have any questions or would like additional information, give us a call at 847-297-1700 and we’ll be in touch with you shortly.