How do I determine which specific assets to pull into the content pool from which recommendations are made?
Do not focus on analyzing data and insights around the buyer's journey on any given page because a buyer's journey is far from linear.
Work at ensuring that the content pools from which you are making recommendations represent a mix of varying journey stages, such as TOFU, MOFU, and BOFU. Each stage is based on the context of the page or site section. The buyer is empowered to pick their journey, and you are well-positioned to nudge the visitors down the funnel. You should recommend the content that best maps to the buyer's interests from each pool.
For example, insert a dynamic blade at the bottom of a page and a single recommendation on the right rail. On the bottom of the page, you make a recommendation from a content pool comprised purely of TOFU thought leadership content. On the right rail, however, you recommend MOFU content to nurture your audience and to provide opportunities to nudge visitors down the funnel.
How do I improve resource page recommendations to yield more form submissions?
If your pages have little text, you may create a light topic cloud for your assets, meaning personalized recommendations have to use fewer keywords. While those keywords might be appropriate, they might not be well-personalized. A light topic cloud may also mean light interest enrichment of the individual's interest profile after they consume that page. Try to enrich a page to improve your ability to map the best resource page to the visitor and ensure that the individual's interest profile accurately represents their interests.
With content changes, is SEO helped or hurt?
No impact on SEO. Using dynamic content to replace the static content within the specific CSS selector with new, dynamically generated content variations is considered SEO-friendly. Google's ranking algorithms rely on a page's default, static version – instead of on the potentially unlimited amount of personalized variations a page can have.
Does adding the widget affect page load time?
No. Whenever you add third-party code to your site or web app, add JavaScript to whatever else you already have. Optimizely's JavaScript runs asynchronously; it will not stall the other JavaScript already on the website. The load time of the widget is generally unnoticeable.
Can Optimizely Content Recommendations make recommendations from a pool of gated content only?
Yes.
Can I add case studies, tools, webinars, and events to Optimizely Content Recommendations?
Yes. Case studies and webinars are already sectioned. Upcoming events require constant supervision, and the date/time for events must be removed from the recommendation engine programmatically, so you should not add these to the recommendation engine.
Can I know if someone is using a paid search or organic search?
You can filter by channel in Adobe to get these insights.
Does Google know when AI is used on the site?
As a general rule, recommendations do not work against you. By improving digital experiences, you can improve engagement metrics, increase loyalty, and build a stronger brand. From an SEO perspective, penalties occur when you deliberately attempt to manipulate organic rankings through optimization initiatives targeting specific variations to search engine user agents (like Googlebot) and human visitors.
Googlebot is not targeted with one set of content while showing other content to users. Content Recommendations do not redirect or negatively impact page loading.
You should ensure that content that is important for SEO displays in the static source code and does not rely on the dynamically injected content. This way, most bots, including Googlebot, are exposed to important content for your SEO strategy.
What content should be recommended right after set up?
Optimizely Content Recommendations ingests every piece of content from across your digital properties. When ingesting content, two things happen:
- Content Recommendations make a copy of the visual components of content, and available metadata in the page source code is also captured and stored (title, URL, Image, Publish Date, and so on).
- Content Recommendations applies NLP to the content, which automatically reads and extracts many topics from every piece of content, assigning each content its weighted topic cloud. The objective of machine learning is to automatically create a uniform and consistent taxonomy across content that is granular enough to differentiate every piece of content's topical profile from one another.
User profiles that may be stitched to certain client-persistent unique identifiers available in the browser are built simultaneously. As users engage with content, the content's topic clouds get attributed to the user, creating individual interest profiles. With each subsequent engagement with content, that profile gets recalculated and updated in session, factoring in topic-weighting in content, topic position, recency, and frequency of topic engagement.
When next-best-content recommendations are made, the AI identifies the content that best matches the user's interest profile. The recommendation decision is made from across "all content ingested" or narrowed down and defined (such as "blog content" or content containing a specific UT code).
Can Optimizely Content Recommendations identify the keywords the user typed into Google Search?
Also, can it Identify what search terms they use in the internal search box?
Content Recommendations does not identify keywords that the user searched. Content Recommendations only captures interactions with content pages and aggregates the topics associated with the first-party interactions in real-time when creating the individual interest profile. To the extent that keywords typed into Google search or other search boxes exist as topics on pages identified by Optimizely NLP, they are automatically included in the user's interest profile.
Please sign in to leave a comment.