Optimizely Content Recommendations

  • Updated

Optimizely Content Recommendations anonymously tracks visitor activity on a website to build a profile for each visitor. Analysis of a unique profile lets you deliver the most relevant content to each visitor.

For example, based on the visitor's activity on the website, the topic cloud in the following image shows a visitor's interest in retirement, which you can use to recommend articles, blog posts, or other content related to retirement planning

Image: Interest profile topic cloud

Content Recommendations uses Natural Language Processing (NLP) to match the content that shares the greatest similarity of an individual user's interest profile in real-time. The corresponding recommendations change appropriately as the user's interest profile changes with more content consumption.

Content Recommendations has many evaluation tools to help you improve your content. For example, the topic performance chart shows how much interest there is in a topic. See Topic performance.

Image: Topic performance chart

You can use Content Recommendations with the rest of Optimizely in the following example ways:

  • A/B Testing – Use built-in A/B testing with your recommendation blocks to determine where a block gets the best results on the page. You can monitor a block's performance over the life of your A/B test to determine which placement or design gets the best click-thru results.
  • Visitor Groups – Are you trying to decide which persona should see which content? Create a visitor group on a particular page, and use two different blocks and deliveries to deliver two different experiences. You could deliver content-group-A to new users to the site, which selects from a different content flow and section, then content-group-B to returning visitors, depending on your tactics.


The following terms are associated with Content Recommendations.

  • Content dashboard – Lets you analyze the number of content items, unique topics, the average number of topics per content item, natural language processing, and so on. To use dashboard filters, see Filtering. The Content Dashboard also lets you analyze the following:
    • Sources – Shows the URLs where content items are viewed and interacted with.
    • Contents – Shows each content item processed by Content Recommendations.
    • Sections – Shows groups of content based on flow rules.
    • Flows – Shows rules for sections. You also can create rules.
    • Properties – Shows the Internet and email areas that Content Recommendations tracks.
  • Insight Dashboard – Lets you track and analyze visitor interactions with your site and the following:
    • Topic Performance – Shows a graph of Volume versus Uniques and plots topics on the graph.
    • Content Utilization – Shows content that performs well or not so well.
    • Profiles – Shows details about a visitor profile.
    • Goals – Shows conversions.
  • Engage dashboard – Lets you analyze recommendations and the following.
    • Deliveries – Set up deliveries to drive personalized marketing to key customers.

      The menu item Deliveries requires extended user rights. Contact your Content Recommendations administrator.

  • Settings – Lets you configure the Content Dashboard.
    • Configuration – Set up master filters or IP address filters.
    • Shared Views – Shows views shared by Content Dashboard users.
    • Topic Selections – Displays private or shared selections.
  • Source – The origin of content that is ingested into the system.
  • Content item – A web page that has an identifier associated with it.

    A content item can contain many topics. For example, if you use a filter to look for insurance, Content Recommendations finds insurance on pages and returns the URLs where it is found.

    Content Recommendations analyzes and audits content to show the following for each content item:

    • source
    • URL
    • import and publish dates
    • text content
    • metadata associated with the content
    • images linked to the content
  • Topic – An indexed subject, such as finance, insurance, hamburgers, rocks, or brand name.

    A topic is prioritized (weighted) based on the following criteria:

    • frequency in a content article (higher weight value)
    • frequency across content articles (lower weight value).

      If a topic appears many times in an article but also appears in many other documents, the weight is the average of the higher and lower weights.

    • position of a topic in a content article
    • time (how recent) since interacting with the topic. A topic becomes less weighted if a user has not interacted with content articles that include the given topic for a longer time.
  • Section – Groups of content based on content or asset type, business category, industry, language, or any grouping criteria of your choice.
  • Ingestion – The process of gathering topics for analysis. For example, if a URL is added to your website where Content Recommendations is implemented, that content is ingested into the system. If content changes after it is ingested, you can reprocess the content.
  • Flow – Rules for sections based on titles, URLs, or metadata to group content.
  • Goal – A set of behaviors you want someone to perform, such as filling out a form, requesting demos, downloading a resource, or viewing three pages on the Financial site. This helps to determine how interested a visitor is in the content. The idea is that people who view three pages on a financial site may be interested in, say, Retirement planning.
  • Session – The period of activity for a visitor.
  • Unique – A specific visitor to your website for a session.
    A new session occurs if a visitor visits your website after a previous session expires and adds to the unique visitor count.
  • Volume – The number of times a unique visitor views pages on your website during a session.