Knowledge base

Where answers meet questions

Personalization Personalization

Traqli Personalization helps publishers deliver personalized recommendations to their audience via email and on websites. It recommends stories based on behavior and content preferences of each visitor. 

Using one line of JavaScript code on the website we analyze what, when and how users read, and then, by applying semantic analysis we identify the meaning of each article or product description on the website. That combined with machine-learning technology empowers us to pick the most relevant stories for individual website audience member and bring him back to website day after day.

Semantic analysis and languages support Semantic analysis and languages support

In order to identify individual users interests Traqli performs semantic analysis for each content piece that have been published, so in other words we can identify the meaning of your content in an automated way.

Here is a list of languages you can start working with right away: arabic, armenian, basque, brazilian, bulgarian, catalan, cjk, czech, danish, dutch, english, finnish, french, galician, german, greek, hindi, hungarian, indonesian, irish, italian, latvian, lithuanian, norwegian, persian, portuguese, romanian, russian, sorani, spanish, swedish, turkish, thai, bulgarian, estonian, macedonian, polish, slovak, slovene, serbian, ukrainian, tamil, bengali.

Traqli currently supports over 40 languages and we are able to add support for many specific languages on-demand, so please send us a request if you didn't find your language in the list.

How Traqli selects the right content How Traqli selects the right content

Traqli captures cookies of every visitor, so they get new stories accordingly to their behavior and content preferences. To make content recommendations accurate and relevant Traqli analyzes:

  • what kind of content visitors read;

  • how much time they spend on each article;

  • the meaning of the content visitors read (using semantic analysis engine);

  • groups behavior patterns;

  • the meaning of new content gathered through RSS feeds.

When we have all these information we can find the right conformity for every user and provide relevant content for her.

Leveraging Traqli’s personalization and content selection approaches you can combine any type of content (e.g. articles, products, advertising) and any type of recommendation approaches (personally recommended, trending, latest or ordered) in your emails or onsite recommendations in a completely automated way.

Recommended content is the most relevant for each user and this type of content selection helps to increase audience engagement better comparing to other approaches. But at the same time the combination of most popular and personally recommended content has the best balance, since personally recommended stories provide the most interesting content related to recent user interests while top trending content has a great discovery function, so users can find something they never paid attention to before.

It’s important to mention that we send only unique and new content (not read by the user). There is rare case when the user can get recommended content that she already read, it can happen in case if user reads the content on a different device from the one she uses to read email (which means there is no connection on a cookie level between browsing history and email address).  Also, in case of emails, Traqli finds new content for every subscriber from the moment of last sent email. When there is no user behavior data (when you upload existing data base) Traqli starts with most read content selection for unknown users, since this type of content provides higher engagement comparing to just latest stories selection, but once the user make a first click next email will contain personalized content selection.

Targeted ads in emails Targeted ads in emails

We’ve developed a solution driven by machine learning that brings geo and device targeting of advertising into emails which we call Dynamic ads and the main idea is to select content at the moment of email opening.

One of the key priorities for publishing industry is effective monetization. And to address this need Traqli can offer contextually targeted native ads within email campaigns. Traqli’s personalization engine driven by machine learning technology that is using 1st party data about audience (based on user behavior and content semantic analysis) and applies it to 3rd party content from external native ads networks to deliver contextually, geo and device targeted native ads.

Here is how it works: when user opens an email there is unique dynamic link inside the template the sends a request to Traqli transfering location, email and device type data to the server, based on the knowledge about this subscriber and geo and device restrictions, Traqli selects the most relevant content from the ad networks. On the next step Traqli, based on predefined layout, merge image and title of the content item into a single image and sends to the email, so technically the user sees a single image which looks like a regular content item (image + title) and contextually targeted to her interests.

As a result you can increase monetization effectiveness and get a new revenue stream which doesn’t face ad blocking problem and provide a great user experience.


Integration code Integration code

For using all benefits and opportunities of Traqli, you must add integration code to your website. You can find the integration code in Traqli Dashboard in Integrations section. You should put the code on every page of the website to enable Traqli see what content does your audience like.

 The integration code is responsible for three functions:

  • analyzing user’s behavior about what, when and how they read;
  • showing opt-in forms;
  • showing on-site recommendation widgets on the websites.

 Note: This code doesn’t slow down website’s work and it loading asynchronously.

Delivery time personalization Delivery time personalization

Traqli personalizes delivery time of email newsletter for each subscriber in the context of previous clicks that user made from the newsletter within the time frame you set. For example, the user clicks on the links at 9 am, the next emails will start coming at 8:30 am.

If we do not have any data about the user, the delivery time of email newsletter is evenly distributed.