Personalized Content – Why It Matters so Much to Your Visitors
Personalization technology is one of the most important marketing decisions you can implement. Offering personalized content recommendations to your website visitors increases your ability to offer relevant content to known and anonymous visitors which increases the likely hood they’ll engage with that content and convert.
When most people visit your site, they aren’t ready to invest their money right away. They want to do a sort of test drive first. So, they visit your site with a research mentality to absorb information about your company and what you have to offer.
By providing content that fits their needs, you encourage content engagement and content consumption. That, in turn, can lead to higher qualified leads.
Artificial Intelligence (AI)
To provide the personalized experience your leads and visitors expect, Hushly’s AI engine is constantly trained on three sets of data to ensure that its recommendations are current and accurate.
The first dataset is the visitor information on your own website. As an example, every time someone views a piece of content, engages with it by spending time on it, navigates through a series of content assets, converts to lead, all the interaction data along with the visitor’s session information, demographic, and firmographic information is made available to the AI engine.
The second dataset is the actual text of the content assets on your website. Using Natural Language Processing (NLP), the content is parsed and consumed by the AI engine ,
The third dataset is the intent data that we pull from our third-party partners that provides information on what your web visitors browsed on third-party websites prior to landing on your website.
These three datasets provide a comprehensive knowledge base for Hushly AI to train on.
Hushly’s AI Engine is hybrid in nature. What that means is that it combines recommendations from multiple models, each of which focuses on a specific method of recommendation. The recommendations are fed into our arbitration engine which determines the best set of assets to deliver. These models can be categorized into four high level types
Content Similarity Models
These models are trained using unsupervised machine-learning method. They use natural language processing (NLP) to read the content and metadata of assets and group them into similar topics.
Using this type of model, Hushly AI recommends assets that are similar in content to the assets already viewed by a web visitor. For example, if a web visitor views content that uses terms specific to non-profits, these models recommend other content assets that are also related to non-profits
Collaborative Filtering Models
The collaborative filtering algorithm is where things really start to get interesting. These models are trained using a supervised machine-learning method. They use the interaction history and visitor information of all the past visitors to form connections between content assets for similar users.
Using this model, Hushly AI recommends pieces of content that were viewed by other similar visitors who also viewed the content viewed by the web visitor. It’s not as confusing as it sounds. Think of Amazon with its “shoppers who viewed item X eventually bought items Y and Z.”
Session-based Similarity Models
This is a deep-learning model trained using recurrent neural network (RNN) methodology on the entire session history of past visitors. By modeling the whole session, our AI engine can provide super accurate recommendations
Using this model, Hushly AI recommends pieces of content that were viewed by users who took a similar path in viewing content assets as the current web visitor.
These models consider the popularity of content assets across all historical visitors on your website.
Using these models, Hushly AI recommends trending content assets as well as most popular content assets to a web visitor. These models offer useful recommendations for a brand-new visitor for whom no interaction history is available to provide personalized recommendations.
Intent & Firmographic Triggers: The Secret to Creating Personalized Content
- Buying Stage
- Profile Fit
- Topic Cluster
- Profile Score
- Intent Score
(Source: Bombora, 6Sense, etc)
- Buying Stage
- Revenue Range
- Existing Customer
- Company Name
- Company Size
- Company Domain
- Paid Traffic (LinkedIn, GoogleAds, Facebook, Twitter, etc)
- Page URL or URL Path
- Referrer URL
- IP Address
- Device Type
- Geo (Country, State, City, Region)
Add Even More Value by Providing Personalized, Related Content to Your Customers
Your goal is to ensure that your content is available to the right visitor at the right time, all while ensuring that their experienceis exceptional.