Labels are a single word or a multiple word phrase you can add to the default language version of a specific article in Help Center. You can use labels to influence article search relevance, to influence Answer Bot results (if you are using Answer Bot), or to create a list of related articles based on labels.
You can add and remove labels on individual articles or you can change labels on multiple articles at once. You must have Guide Professional or Enterprise to use article labels.
Adding and removing labels on individual articles
Labels are a single word or a multiple word phrase you can add to the default language version of an article in Help Center. You can add articles labels for multiple purposes, including to influence search and Answer Bot results or to create article lists (see Understanding why to use labels).
You can add and remove labels on individual articles or you can change labels in bulk for multiple articles at once.
To add or remove labels to an article
-
In Help Center or Guide, navigate to the article where you want to change labels, then click Edit article in the top menu bar.
-
(Guide Enterprise only) If you are on Guide Enterprise, click Article settings. This step is not necessary on Guide Professional.
-
In Labels in the right sidebar, start typing the label you want to add, then select Add as new label or select the matching label, if it exists.
You might have to scroll down to see labels. See Best practices for adding labels below.
Labels live on the default language article and not on translations of the article. If you have translations, you can add labels in multiple languages to the default article.
-
Add multiple labels as needed.
-
If you need to remove a label, click the x beside the label name.
-
Click Save.
You must manage labels on each individual article or using the API. There is no global management for article labels.
Changing article labels in bulk on multiple articles
You can add or remove labels to on multiple articles at once.
-
In Guide, click the Manage articles () icon in the sidebar.
-
Find the articles you want to update by browsing, searching, or using a new or existing articles list.
-
Select one or more articles to change labels.
You can select a maximum of 30 articles at a time.
-
Click the Article settings menu at the bottom, then select Change labels.
-
Do any of the following:
- Add a new label: Enter the label, then click Add as a new label.
- Add an existing label: Search or browse to find the labels you want to add, then select any empty checkbox or any checkbox with a minus.
The minus sign () indicates that the label appears on some but not all of the selected articles. When you select it, it changes to a checkmark (), indicating that it will be added to any selected article that doesn’t already have that label.
You can select multiple labels to add to the selected articles.
- Remove an existing label: Search or browse to find the labels you want to remove, then deselect one or more labels.
The checkmark () indicates that the label appears on all of the selected articles. The minus sign () indicates that the label appears on some but not all of the selected articles. When you select a checkbox with a minus sign, it changes to a checkmark.; click it again to deselect the checkbox.
You can select multiple labels to add to the selected articles.
-
Click Change.
The label changes are applied to the selected articles.
Understanding why to use labels
Labels have the following uses and impacts:
-
Influence article search relevance
Adding labels can make your articles more search friendly. They are not designed to be used for fine-grained control of ranking. For example, if I have two articles about making waffles, one with the word “temperature” in the body of the article, one with the label “temperature”, the labeled article will be ranked higher if a user searches for the word “temperature”.
Labels are indexed for search with a bit less weight than the article title, but multiple labels with similar words can outweigh the title and body of the article. Use labels carefully as you can significantly impact the relevance of your search results and usefulness. If you are considering using labels to impact search relevance, consider that efforts to do so could end up with inferior ranking performance for your users. The best way to have your article perform well in search is to create a short and focused article, where the title and body include keywords and clearly connect to each other.
To learn more about search, see How end-user search works in Help Center.
-
Influence Answer Bot results
Answer Bot uses labels to influence the results it shows. It allows you to define a whitelist of articles that are allowed to be used when looking for relevant matching results. You are able to add up to ten labels that serve as an OR filter, allowing you to expand the whitelist to any articles containing any of the defined labels.
To learn more about Answer Bot, see About Answer Bot.
-
Create article lists
Article lists enable you to get an overview of all your published and unpublished knowledge base content, and then refine that view by using search and applying filters to build article lists. For example, you can find articles that have a specific article label, such as out-of-date.
To learn more about article lists, see Creating article lists to different views of your knowledge base content.
Best practices for adding labels
Labels can help boost the search relevance of an article. However, you should use labels carefully and sparingly. It’s more important to make sure the article title and body contain the relevant keywords.
-
Use single word labels where possible, instead of multi-word phrases
It is possible to add labels as single words or as multiple words or phrases. In general, it’s more efficient to use single word labels. If you add a multi-word label, the search engine breaks it into individual words to perform the search. For example, if you have a label of “late delivery,” it gets broken down into “late” and “delivery” for search.
Avoid using long phrases as labels to boost an article’s ranking with respect to a query. For example, “Can I return something I ordered online to my local store.” Instead, you should modify the article’s title or content to make it literally relevant to the query.
-
Do not include variations of the same word, including different tenses or plural forms
You do not need to include multiple labels for variations of a word. For example you do not need a label for “return” and “returns” or “update” and “updated.” Search stemming allows different forms of the same word to match. In particular, the singular and plural forms of a word will generally match.
-
Use a limited number of labels, instead of overloading an article with labels
Use labels sparingly. Adding lots of labels might actually diminish any matches on labels. This is because it is assumed that matches with a fewer number of labels beats matches with more labels. And too many labels might outweigh the relevance of the title and body.
For example, if article 1 has the labels “car,” “automobile,” and “transport” and article 2 has only the label “car,” all other things equal, if the end-user searches for “car,” the article that has only the label “car” will rank higher. That is because, as a general principle, an article that is about one specific thing is more relevant than an article about many things, when a user is looking for that one specific thing.
The number of labels for search is limited to 200. The search engine will use the first 200 labels and ignore any additional labels.
Your best bet is to look at the top ranked search queries and make sure that they exist in either (but not both) the title or the labels. You don’t give content an extra boost if you match a term across the title, body, labels, and comments.
Using the Help Center API to filter articles by labels
There are two ways you can filter articles by labels using the Help Center API:
-
Article:
/api/v2/help_center/en-us/articles.json?label_names=foo
-
Search:
/api/v2/help_center/en-us/articles/search.json?label_names=foo
These two endpoints behave differently and will return different article results. They are programmatically different and query different data sources for their results.
Comments
0 comments
Please sign in to leave a comment.