First of all, what is structured data? Structured data is the addition of basic data (film, cooking recipe, product, person, event, company, rental, etc.) and a specific structuring scheme, implemented in the source code of your Web page. Structured basic data allows Google to better understand your content…


What is a structuring scheme or markup scheme?

In 2011, the 4 major engines, Google, Yahoo, Yandex and Bing, agreed on a common “lexicon” for structuring basic data (film, person, event, etc.). They then create the site which brings together all the existing schemas makes them evolve and creates new ones.

So we now know that a film is a given
basic and that can help us structure this basic data, if
the scheme of course exists.

We will see a little later what these diagrams look like
markup and how to use it to structure your data.

Semantics make sense

Search engines are just aggregators of
data, these are just machines that have been using the same methods for
the beginning.

Synthetically, an exploration robot (crawler, bot)
browse pages by following the links found there, each page is
stored, analyzed, indexed and restored in the search engine if the
content is compatible with your request.

This operation, summarized in its simple use, did not give
not necessarily make sense of the results. Over time, analysis techniques
contents have improved to integrate the notion of meaning: what
my article is talking about, what my site is talking about, what the user is looking for.

The Hummingbird algorithm (Colibri)
launched in 2013 opened the doors of the semantic web by modifying the approach of
engine vis-à-vis the requests of Internet users.

Hummingbird (Colibri)
allows the engine to understand the request, the direction of the request; it turns
always based on words, but tries to detect the desired result
by the user by analyzing the association of the words present in the request, it
can enrich the results presented by the association of synonyms or the
replacement of words in the request.

We are therefore no longer on a “strict keyword” result,
but on a result corresponding to the direction of the request, which means that
results may appear when content does not use a term

It should be noted that we are on so-called requests from
long tail and close to natural language. Hummingbird (Hummingbird) can only be effective if the request is
long enough to allow this exercise in understanding,

Why should I structure or tag my basic data?

Structured data uses so-called tags
semantics, and therefore allow the engine to improve its understanding of
data it needs to process.

Structured data specifies whether we are talking about a place,
of a person, an organization, a product, an event …

We contextualize our basic data, we frame it
of meaning and offer the engine a simplified understanding of our content. In
making our contents digestible, we are approaching the request for
the user.

In addition, tagging basic data can provide
enriched data to the engine and therefore to modify the display presented in the

Result displayed for a recipe
kitchen with structured data “RECIPE”. The RECIPE scheme
is the diagram dedicated to cooking recipes.

Result displayed for a recipe
without structured data.

Let’s take another example:
the one on a page that talks about and sells a book. On the page of
search engine results, in addition to Title and Meta Description,
the following elements may appear: the price of the book, the evaluations
of readers who bought the book, a photo of the cover, etc.

Structured content coupled with quality writing is necessarily

To note :

  • Applying structured data does not always guarantee an enriched display, the engine being the only decision maker;
  • Structured data does not boost positioning, but sending more understandable content to Google, making it more relevant and therefore more in line with user requests, will certainly lead to more impressions. And if we have, in addition, a rich extract (rich snippet) in the serps, this should help us to improve the CTR (click rate).

How should I structure my basic data?

Basically, the aggregation of engine data is just one
sequence of words, if we find in a content the following sentence “Mike
Jagger is going to perform at the Empire State Building “, we don’t know that Mike
Jagger is a person, that the Empire State Building is a place.

By extension or relationship, we could offer thanks to
structured data from additional sources of information, contextualizing
at best this person or this place:

  • Mike Jagger is a singer, songwriter, musician. He founded the group The Rolling Stones. He is British, he is a man, he has children, he has a wife, etc. ;
  • Empire State Building is a building built in 1930, located in New York, in the state of New York, it belongs to an organization or a person, etc.

We could go very far in relationships

We therefore base ourselves on the bible of structured data
what is lists all the existing possibilities of
structured data markup.

There are 3 formats for encoding this semantic markup.

  • MicroData and RFDa: are markup that we affix in the HTML code;
  • JSON-LD: is a script that we add between the tagsof our page (s) or page templates.

Example Type
Movie for the text “Ghostbusters was produced by Black Rhino in
the United States ”.

MicroData code

RFDa code

JSON-LD code

We will only be interested in the Json-LD format, the W3C (World
Wild Web Consortium) considers this format as the new standard to use.

Let’s detail the proposed Json-LD script:

This brace corresponds to the
closing the opening brace before @context, without comma, because
we don’t add anything
We close the script

Make sure you haven’t made a mistake using Google’s verification tool by copying and pasting your code or testing your URL.

Click on test

It is interesting to note that the auditor
request to complete the information with fields it associates
usually to the type chosen, it asks to add a visual, a date of
creation and a director.

Note also the difference in symbols: the lack
image is considered an error, the other 2 as warnings.

We will therefore modify the code accordingly. Go to your movie page to know the exact syntax of the properties to add.

We discover that director corresponds to the type @person.

So we introduce @person nesting to our type

Remember to open the {and close them} at the right time.

Again let’s test our code.

The warning on director
is fixed, we will complete the code so that we no longer have
error or warning.

With the addition of the image, we have a “PREVIEW” button
that appeared, allowing us to visualize the enriched result.

We could add additional information to this
code by integrating the actors, a description, etc.

The use of the test tool is important because it
allows you to validate or not your code.

Very long scripts can be subject to error: addition
comma in wrong place, lack of comma, opening brace problem
and / or closing.

The same code with a comma too many after the name of the

I invite you to use a text editor, such as
sublim text or other, with syntax highlighting to prepare your scripts and
work with the indentation which highlights the discrepancies between
tags, to visualize your nestings.

But how do I do if I have thousands of films or products to tag
for example ?

This scenario is quite simple to solve, you connect
your json-ld to your database, you most certainly already do
with OG tags (open graph).

But what should I do if I want to include information that I don’t have
not in the database?

It all depends on the type of data you want
add, but you will most certainly need to use type databases
open data and recover data via APIs.

Open data databases, are databases
open, accessible to all and free of rights.

You will find open data databases in different sectors: transport, culture, weather, finance…. I advise you to read this article.

What can I structure as data?

If you use the list, you don’t have
that spoiled for choice

The most used :

  • Organization: which provides information about the company;
  • Person: which describes a person;
  • Website: which gives information about your site;
  • Breadcrumbs: for a thread of breadweed in serps;
  • Product: for enhanced display in serps (image, price, etc.);
  • Aggregate and review: for managing stars and reviews;
  • But also restaurants, events, music, films, local businesses, jobs, articles …

How do I go about structuring my data?

It’s a job that can take a long time to prepare.

If I were to offer you a methodology: “Let’s go
from general to precise. ”

Imagine that you have an e-commerce site.

  • Structured data: organization, website, breadcrumbs (general);
  • Structured data template, product listing: itemlist (precise);
  • Structured product data: Product (precise) + opinion + notes.

Each structured data can contain little or a lot

Organization can be this:

Or this :

I invite you to identify your templates and for each
template, to think about what you could structure as data.

The point of tagging is not to tag everything, but to
tag information that is relevant to the user and to Google.

Read the Google guidelines on the subject.

Who can help me structure my data?

Learning the structure of a Json-LD script is the
best way to structure your data with as much information as possible
to send. Mastering nesting opens up a multitude of relationships between types.

But alternative solutions exist:

Useful links

About the Author :

Corinne Briche has been director of the SEOh agency since 2015. She has been evolving in the digital world for many years and has worked as a Senior SEO consultant in agencies as well as with the advertiser on complex SEO issues with high stakes for companies.