SEO e-Newsletter Issue 36 – 2021.07.01~2021.07.15

Summary - Recent algorithm updates

Since April, there have been many "official announcements" of algorithm updates. Those have been fully launched and those are still in progress. Here is a small summary to help everyone clarify.

As shown in the table, many algorithm updates are not released all at once, but are released over several days or even weeks. Therefore, if there has been a recent change in website traffic, it is most likely not caused by a single reason. These updated information are shared in e-newsletters and Facebook. If you are not sure about any item in particular, please leave a message!

[Interesting Fact] Google confused WordPress.com & WordPress.org

in An interesting case seen ontwitter, Google mistakenly confused WordPress.com with WordPress and rewrote its meta description so that.org'sDownload pageappeared.comThere are so many great words, which may greatly mislead users who are not clear about the difference between the two!

Just a quick introduction, WordPress is an open source technology..orgis maintained behind the scenesNon-Profit Organization, while.comuses this technologyFor-profit organization. This difference may be confusing for people who come into contact with it for the first time.

Many people know that Google will rewrite the meta description of a web page to make it more consistent with the user's search intent (or search terms). The general advice is that when it is rewritten, you should go back and examine why the original description was judged to be less relevant and rewritten, and rewrite a better version; but in this case where it is obviously outsourced by Google, SEOs can usedata-nosnippetThis tag prevents Google from using specific sentences in snippets.

Website server relocation checklist

JR Oakes wrote this article about a checklist when moving a website server. The server moving itself is not something I often encounter, but the process involved is something I often encounter after I move to inhouse. After reading this, I will have a more comprehensive understanding of the process behind when the website is updated.

This process is roughly divided into four stages: 1.) preparation period, 2.) test environment (staging environment), 3.) moving, 4.) after moving. The article lists things to pay attention to at each stage. Although SEO personnel are unlikely to be engineers who perform these steps, they are usually a part of QA. It is recommended that everyone understand their role in this process regardless of whether they are exposed to it in daily work.

link:Server relocation checklist

2021 SEO Practitioners Salary Survey

This is a survey conducted by Search Engine Journal. The regional distribution of the responding groups is 61% in the United States, 3.8% in the United Kingdom, and 2.4% in Canada. This is just a preview at the moment and the full report is expected to be released in the first week of September. Although the respondents are mainly from the United States and have different national conditions, there are several interesting points:

• The salary of SEO freelancers is lower than that of other groups (full-time employees), but this may be because 12% of the researchers are from India, which has lowered the overall average in the current prevalence of SEO outsourcing.
• The salary of the company founder is more likely to exceed $200,000/year, which is 104% higher than Director & C-Level SEOs

link:SEJ Article

Google Search VP talks about MUM’s long-term development

MUM (Multitask Unified Model) is a new milestone in machine learning by Google. It was unveiled at Google I/O this year and its capabilities were officially evaluated.BERTis 1000 times stronger, in the 32nd issue of Digital EngineNewsletterwas introduced before.

Pandu Nayak, VP of Google's search department, was interviewed to discuss the short-term and long-term development of MUM. First of all, he said that MUM will not be just a one-time product update. Many departments in Google are studying how to implement MUM into products. Currently,Best known applicationuses MUM to learn more than 800 different names for the COVID-19 vaccine in 50 different languages around the world.

In the medium term, they hope to make practical use of MUM's multi-modal features (across text, images, videos...). For example, users can take a photo of shoes, then upload it and ask Google "Can I wear it to climb Mount Fuji?」。

The long-term goal is actually to return to the point of better understanding the user's search intentions. Nowadays, many relatively long sentences (search words) cannot be "understood" by Google; the other is cross-language applications. Pandu is inIn another articlehas a good example --想像你會中文和英文，看了一本英文書，再用中文告訴朋友書在說些什麼，你能做到這樣是因為你理解了書中的"資訊"，而這個"資訊"是不受語言限制的。

link:SEJ interviews Pandu / Google I/O MUM Video / Digital Engine Facebook Introduction

Backlinko analyzes 3.6 billion articles to understand which ones are most effective over the long term

inThe 30th issue of the e-newsletteronce introduced a research report by Buzzsumo, pointing out what kind of article title is attractive enough. This time Backlinko (founded by Brian Dean) launched another study to analyze which articles are most successful in the long term.

In the field of English content marketing, it can be roughly divided into articles with longer validity periods.everygreen content, or for short-term hot topicstopical content. There is no absolute better or worse between the two, and this analysis is to point out the common characteristics of successful everygreen content.

link:Backlinko original text

Search Atlas - Compare search results from different regions (in closed beta)

I accidentally discovered the very interesting research project Search Atlas. Everyone knows that Google will return different results according to the searcher's region, but the researchers want to create a platform to compare how different Google search results are in different countries.

For example, the search results for "god" in Japan, India, or Western countries are quite different.

This difference will also be extended to the political level, such as the issue of the Crimean Peninsula. Search results in Russia will say "sovereignty is disputed"; in Ukraine it would be said"embezzlement";while other European countries will focus on"EU sanctions against Russia"

This last analysis can also reflect the differences in how countries with different economic strengths view things. For example, for "climate change", in high-income countries people will discuss "how to protect" or "how to protect climate change"; but in low- and middle-income island countries, there are more search results describing how terrible climate change is or how it threatens livelihoods.

This project has a relatively low relationship with SEO, but after reading it, I found it quite interesting, especially now that many people are beginning to reflect on whether there are any fundamental problems with the way humans absorb information through technology. I would like to share it with you!

link:Search Atlas official website / paper / Register as a trial user

[Official Video] The website was manually punished and what to do next

Google uses algorithms to catch most of the spam, but a small number of them still use manual review, which is the so-called manual actions. In this episode, Aurora discusses manual review, content monetization policy violations, and what to do if a site inadvertently violates these policies.

When a website is determined to have spam, some web pages or even the entire website will be affected. The possible impact is a lower ranking or disappearance from search, depending on the severity of the penalty. There are many types ofSpam, for example, "Too many unnatural links from your site」、 「Structured data contains incorrect information"...etc. are all possible reasons for being punished.

Google has many content policies, and they are slightly different in different places (Google News, Google Discover, structured data...etc). I recommend this articleofficial document, which organizes all search-related policies in one place.

[Official Video] How long does it take for the SEO of a new web page to be effective?

If you know John Mueller a little bit, you can probably guess that his answer is... it depends (not necessarily)!

First of all, inclusion is not guaranteed. Google and other search engines will not index all pages on the Internet. In order for a web page to be included, the search engine must feel that it is worthy of inclusion. Secondly, even if it is included, there is no guarantee that users will see it. The webpage may not rank for any search terms, or it may be removed from the index over time.

If your webpage provides value, then the technical part is to ensure that the server is functioning properly, actively submit URLs, link appropriately to new pages, avoid unnecessary webpages, etc., so that Googlebot can easily crawl your page and include it.

[Abstract] Google Search Podcast – Spam & Gary talks rankings

This episode invited Duy from the Search Quality team, who was also an SEO before joining Google.

Webspam
Duy mentioned that spam is mostly detected through machine learning methods. You can think of this process as training a person to read a web page. It is actually quite easy to determine whether the content above is spam; Google will do this every dayfound 40 billionspam pages, but less than 1% of the search terms will cause these spam pages to surface; now they are facing a more difficult situation: there are too many "old" websites in the entire online world, and some of the technologies or plug-ins used on them may become targets of hackers due to their obsolescence. Google is cooperating with more website hosting platforms to prevent these neglected old websites from becoming a breach of hackers and spam content.

In addition to spam (junk), scam (fraud) is also a big problem. For example, many people will want to rank in "XX company customer service”, fake customer service and then deceive. Since 2018, there have been some algorithm updates to avoid this situation.

Gary talks rankings
The previous episodes of the podcast mentioned a lot of crawling and serving (#25、#27、#30、#32-33、#35), Gary doesn’t want to talk too much about the ranking part. After all, it is Google’s industry secret, so this episode talks about the basic principles of information retrivel (IR).

is a bit of a continuationPrevious issue's discussion, the ranking system has two major stages. One is a simple classification to find the 1000 web pages (or documents) that best match the search terms, and the second is the use of a more complex ranking algorithm (ex: RankBrain) for the final ranking. The basic way that sorting works is to give each document a "score", and this second stage is to give each document a "multiplier" for its score.

For example, if the algorithm determines that a document is highly related to the search term, it will assign a high multiplier to make its final score very high, that is, it will be ranked higher. It is usually difficult for two web pages to have the same score, and some small algorithms (such as HTTPS) are designed to act as tie-brakers.

link:verbatim draft