HTML Tables as Google Rich Snippets — Breakdown, Takeaways, Achieving The Snippet

In late 1993, Dave Raggett proposed the digital table schema markup display of complex material and data, using rolls & columns. In October 1994, W3C was established as a standard, thanks to a collaboration between MIT and CERN, with the help of DARPA and the European Commission.

Fast-forward to the year 2007, and we see the first appearance of additional elements in Google's search results. This practice has been further developed to include now a large variety of rich snippets, such as information regarding flights, sporting events, videos, recipes, movies, news and much more.

For Google, since 2007, HTML tables have been treated as a standalone element worthy of displaying enriched SERP results. But how do you tell a good table from a bad and misleading one? Several years back, Google started experimenting with "machine learning," and it was eventually incorporated into their algorithms. Through efforts of trial and error, the developers have found a way to teach the algorithm how to depict the worthy tables on its own with ever-increasing accuracy. This helped sieve through the billions of HTML tables with bad layouts.

Right now, the purpose of an HTML table can be associated with a certain search query, which is what we see as "featured snippets." Although nowadays CSS provides far superior layout capabilities and where tables are considered a caveman's practice, many websites heavily rely on them to stay competitive.

Essentially, there are 3 different types of HTML tables:

• Horizontal tables;
• Vertical tables;
  
• Fusion tables.
  

Ideally, a good table will have its first row acting as a header, describing the data below it. This is good because it gives search engines something to work with. Just as articles have titles, HTML tables have captions. Depending on the quantity and complexity of the data below the caption, there are special ways how to create different types of tables in HTML that may be used.

So let us examine the code. Using the W3C standard, this is what the skeleton of an HTML table looks like:

<table> — Indicates that the data will be contained within a table instead of a <div>;

<caption> — Tag for HTML table titles. This is read and used by search engines;

<tr> — Table row. A tag for generating horizontal table rows. Each new tag generates an extra horizontal row;

<th> — Used for HTML table headings. A tag for generating bold headings within the data;

<td> — Table data. This is the data container of the table. It can contain all sorts of HTML elements: text, images, lists, other tables, etc.

It is good to understand what each element of the table relates to because it is still in regular HTML. Therefore, it can be manipulated via CSS, which is convenient when designing a responsive table for mobile devices and apps.

So, why not divs, you would ask? Well, through observation of thousands of SERP results that trigger a featured snippet, we can observe an interesting pattern:

HTML tables > ordered lists = unordered lists > divs

And what about using div instead of the table? It appears that Google favors the table element more than lists, and even more so than div containers. This makes sense due to the nature of these tags. Tables are standalone elements that structure, contain, and style data sets. Lists are non-demanding elements that can either be casually used or enumerated. And finally, divs are literally universal containers for pretty much anything.

Before you get excited, we have to destroy your dreams. Unfortunately, there is no such thing as a featured snippet manual that will guarantee you getting 100% result. This was publicly announced by Google and is also mentioned in their documents. Google check rich snippets by a self-sustaining machine learning algorithm, which is good at depicting queries but not entirely accurate.

However, below we will discuss key takeaways that will highly increase your chances of earning that valuable snippet.

The first order of business is to find how to make a google search query trigger a featured snippet. Even if you have done your work flawlessly, following the advice below does not guarantee the trigger.

Observations show that if the first 3-4 search results are all eligible for a featured snippet, then the algorithm will allow it. This automatically means you will have competition, no matter what you do.

Furthermore, the original purpose of rich snippets was to quickly answer a question or give the user access to what he/she is looking for.

That makes sense, right? The first SERP contains the most helpful and highly accurate results for the user. Thus, if a rich snippet preview is to be shown with the intent to facilitate the user, it would have to be taken from those first 10 results. This is true for all types of featured snippets.

Of course, we cannot say for sure. Some snippets move in a matter of weeks. Others take up to 6 months to update. The important thing to recognize is that the frequency is likely to get shorter as Google's algorithms become ever so effective.

Remember back in the days when major updates such as Panda and Penguin were periodic updates, each targeting a specific issue with searches? Right now, those updates are part of the core algorithm. It is expected that Google rich snippets will follow the same path eventually.

Did you know that Google actually has a separate index for HTML tables? Check it out:

https://research.google.com/tables

It looks like this:

We see an index for web tables and fusion tables. It appears that horizontal and vertical tables fall under the category of web tables, whereas hybrid tables are classified as fusion tables.

You will notice that the URL of the index only exists as:

https://research.google.com/tables

It does not look like local results are being supported. Also, it says that this index is experimental.

Although not mandatory, the <caption> tag, as explained earlier, acts as the title of the table. Without this HTML tag for the table, Google will look for the next best thing - the surrounding headings. They can be anything from <h1> to <h4>. <h5> and <h6> are quite rare, as these tags are typically used to entitle very specific portions of content deep down in the page, usually not very semantic with the topic of the page.

By using the <th> tag, you indicate to search engines that your table has order in its data that is topic-related and patterned all the way from the beginning to the end.

You also have to take a look at the role of the table. What is it about? How big is it? Can it be used as a substitute to describe the purpose of the entire page? What is its relation to the topic of the page?

Of course, Google will look to pick only tables that are the best possible answer to the search query. Therefore, breaking down your topics can actually favor you in some cases. For example, we have an article about a very popular inspirational motto: "Improvise. Adapt. Overcome."

Each of those single-word sentences can be treated as a topic of their own. Thus, all of the following search queries can be considered as a schema for featured snippets list or table:

• How to improvise?
• Improvisation techniques
• How to adapt?
• Overcoming obstacles
• Steps for overcoming obstacle

In each of those topics, it is possible to work in a table with <th> fields, such as:

Not much to say here. Obviously, only the highest content quality will be chosen among the many.

Take note that tables on their own mean nothing. The way Google's algorithms work is by looking at the content that surrounds the HTML code for table of contents - headings, paragraphs, images etc. The best practice is to sort of wrap the table around relevant content, like so:

Having similar or exact data in the table as well as somewhere else on that same page creates confusion. Typically, if the algorithm isn't sure what to choose, it will usually choose nothing.

Text written in bold is historically used to highlight important pieces of text or to style headings.

Although still read, tables with weird proportions, irregular symmetry, tables that are too wide or too long on a X x Y axis may be bypassed.

Again, having data with a few characters in one cell and whole sentences in the adjacent cell counts as an irregularity. This is best avoided.

Unfortunately, two or more columns with competing subjects in snippets search results are not yet received well by search engines. Even though they may be legit within the entirety of the table, the algorithms are not yet advanced enough to tell internal subjects apart.

It is not uncommon to see a table containing numerical values. However, lack of clarity about what unit these numbers are expressed in may hinder the table.

Contrary to what you may think, Google does tolerate the use of relevant links within HTML tables, and yes, it does display them in the featured snippet.

Surprise surprise… HTML tables generate yet another type of rich snippet. Above, you can see an example. We searched for "tesla model S". On page 1 we see a lot of results like this one. But on page 2… ?

Actually, this type of rich snippet is triggered by vertical tables and can be found as far back as page 8 or deeper. We know of other microformats that behave similarly, don't we?

• Breadcrumbs;
  
  
• Review ratings;
  
  
• Events.

What makes vertical tables unique is that they contain multiple <tr> table rows with only one <th> and <td> per table row. The result is a loooooooong vertical list. Wikipedia actually uses this table style in its articles.

An interesting takeaway is how longer tables are actually cropped, based on the search query. Google will choose to display up to 6 of the most relevant table rows below the meta description. Yes, you heard me — 6.

Long story short - Google decided that taking into account user feedback has more value than downsides. And this is what Project Owl is. They have modified their SERP interface by adding extra functions.

One of those functions is the ability to user-rate featured snippets.

What!? Really?

Yup, take a look:

Clicking on the "feedback" option now prompts a window with extra user options. Notice how diverse they are? And then some! There is even a field to type in generic user feedback.

So, should you be worried?

Not necessarily. Take a look at what you get after providing input and clicking on "send":

Your responses help to improve the Google Search experience.

Note: Your feedback will not directly influence the ranking of any single page.

This is just another social experiment. If things get out of hand or turn out really bad — it's bye bye birdie.

Back to the feedback. We have on our hands two options:

• marking the snippet as …
  
  
• commenting on the snippet

The first option might be the dangerous one. Those are fixed options, which can be collected algorithmically in order to rate a certain snippet, similar to how it is done with review ratings (average from 0 - 5). Assuming that upon rating a snippet, the algorithm locks onto your IP and prevents you spamming negative options, this is still a knife with two edges, which is the reason why these new changes have not been received well by the communities.

The other manual option for commenting should be directed to Google's manual team, considering that the amount of queries that trigger a snippet are far from overwhelming.

In the end, it is hard to say how all of this will turn out. There are those that respect the competition by recognizing its collective value. And there are those that just do not bother playing by the rules and are happy to spam the featured snippet that you worked for so hard.

Which one are you going to be? Let me know in the comments.