Enhancing KB Articles with metadata for SEO and Social Sharing

[Erich Zirnhelt]

Last week I shared that we'd made our Knowledge Base content publicly accessible, and hinted that there were steps to be taken to make that content more consumable for search engines.

In this second post, I will dive into the details of the HTML metadata (and some microdata) we added to our articles.   These elements both help search engines understand the content, and also enable a richer social sharing experience.

From the outset, one of the recognized benefits for us of going public was allowing search engines to crawl our articles.   Currently, internet searches drive about 500K visits per month to our other content sources (product documentation and communities).   We know some of the content you are looking for is in the KB, so it was obvious that we needed to reveal them in those same searches.

Without doing a thing to our content's metadata, the crawlers would have crawled it once we opened up, but they would have had more trouble in differentiating our articles from each other.   To help them differentiate, we followed some SEO guidelines and best practices, leveraging standards like Open Graph Protocol and schema.org:

• From OGP, we leveraged some of the Article elements
• From schema.org, we selected from the Technical Article object
• We also took advantage of Twitter's Twittercard standard

Though we added more than I'm highlighting below (e.g., URLs, descriptions, dates, images), these are the ones I felt worthy of some explanation. (For a great summary of the many attributes you can use, check out this post from moz.com.)

Title

Search engines infer a lot from the HTML title of the article.   In our case, the HTML title on every page in HI was set to "ServiceNow Customer Service System". We therefore made sure that we were providing a title string that correctly reflected our article's content.   We also prefixed it with "ServiceNow KB" and suffixed it with the article's ID:

<title>ServiceNow KB: ODBC Troubleshooting Checklist (KB0538995)</title>

There are several title variants supported by the standards, and we addressed those we felt most relevant:

<meta content="ServiceNow KB: ODBC Troubleshooting Checklist (KB0538995)" property="og:title">

<meta content="ServiceNow KB: ODBC Troubleshooting Checklist (KB0538995)" name="twitter:title">

Note: Specifying the title does not guarantee how it may appear in search results.   For example, Google has indicated that they will adjust the appearance of the title to better reflect the content as it applies to the search string.

Keywords

It's fairly common knowledge that search engines give keywords far less weight than they used to, and normally I'd not recommend their use.   However, there are two additional points I'd make here:

1. if you plan of having your content crawled by a local, internal search engine, keywords can be very useful in identifying clusters and auto-classifying articles
2. there are several keywords-like attributes that are still valuable, still respected by the search engines, or used by social sharing sites

Here are the keyword and keyword-like attributes we used, which we pulled from the article's Topic and Category values:

<meta content="General,Integration Solutions" name="keywords">

<meta content="General" name="category">

<meta content="General" property="article:section">

<meta content="Integration Solutions" property="article:tag">

Ratings

One of my favorite elements from schema.org is the Aggregate Rating.   This is what makes the article star ratings visible in search results, and is easy to harness as we have article ratings being tracked already.

Here's how it appears in the HTML.

<div itemscope="" itemtype="http://schema.org/AggregateRating" itemprop="aggregateRating">

<meta content="4.0" itemprop="ratingValue">

<meta content="2.0" itemprop="ratingCount">

</div>

We had to deviate from the standard a bit because our current unauthenticated view of articles does not reveal the rating dialog.   Ideally, it would be blended with that part of the page.

Here's how it ends up appearing in search results:

Validating the Content

There are some great tools available that allowed us to be certain the content was properly marked up.   There's a more complete list at the bottom of the OGP home page, but these are the ones we used most:

• Google's Structured Data Testing Tool
• Facebook's Debugger
• Twitter's Twittercard Validator

Next Steps

We are currently monitoring the health of internet search indexes using some very powerful tools the engines make available (that I plan to discuss in my next post), and may make adjustments from our findings there.   Longer term, we want to leverage more of the schema.org structure at the content level, but that requires some enhancements to our authoring process.