• Setting Personyze Recommendation engine

    A Quick Introduction to Recommendations:

    Recommendation is about presenting (floating out) to visitors the most relevant products, articles and promotional content that will strongly appeal to them, based on the personal information Personyze has gathered from their:

    • Demographics and geolocaiton

    • Apparent interests

    • Page products/articles/videos interactions

    • Container data

    • Referral sources

    And combining it with public or ‘wisdom of the crowd’ data, collected from the behaviors and preferences of other audience members with similar tastes and intentions.


    Please note that:

    • The recommendation settings that are used here can also be leveraged to personalize and send out automatic (triggered) emails.

    • A single page can have multiple recommendation widgets. If it is an ecommerce page, one kind of widget can recommend the best upsells for the particular product category, another widget can recommend the best cross-sells and a third can talk about the most frequently purchased accessories.


    What Do You Need to Get Recommendations Up & Running:

    You must synchronize (upload a file, use Personyze API, or Set Personyze robot to crawl your site) and keep your product catalog updated, the interests that you want to use for targeting certain types of recommendations, a list of the articles and news updates that you want to suggest, and so on. You can use our ‘Managed Services’ option to allow us to take over this responsibility.

    How to Upload Products/Content/Interest List:

    You can manually upload a .csv sheet or you can use Personyze API to call in the data from your company database, or use both.

    Screen Shot 2017-04-25 at 7.48.15 PM


    Why Do We Need Products Catalogue/Content List to Ensure Recommendations Work Properly?

    The recommendations Personyze makes are based on complicated and ever-evolving algorithms. We use the product data (like its name, price, popularity with certain segments) as a part of this algorithm to generate accurate suggestions. The system also pulls Product images and content introduction snippets to entice buyers/readers from these catalogs and lists.

    What Type of Data Should Your Catalogs and Lists Have?

    In addition to the default data such as text, images, and price, we also need (recommended)  information, such as:

    • Similar products/content in catalog for a given item/article

    • The best accessories for e-commerce and best ‘To Know More, Read’ suggestions for content

    • Season-based information like which products are sold the most during a particular season, or content that generates most interest during recurring events like sports tournaments or elections.

    • Availability of e-commerce products

    • Discounts around the corner

    …and anything else that you have in relation to building the right recommendations. The more Personyze has to work with, the more it will learn and improve.


    What are ‘Interests’?

    Interests apply to all types of websites, ranging from e-commerce and content curation platforms to news sites and educational portals.

    We define ‘Interests’ as the broad categories that visitors find stimulating and desirable. And we use multiple ‘interests’ to match them with the most accurate recommendations possible. One advantage of ‘Interests’ is that it is not as narrow and restricting as products. For example if an audience member looks for a tuxedo, we do not set the ‘interest’ to Tuxedos. We set it to ‘Events’ so that we can pitch multiple accessories, cross sells and upsells like ties, wallets, shoes for all events and even car supplies like waxes – everything needed to make an impression at posh events.

    For example: if an audience member looks for a tuxedo, we do not set the ‘interest’ to Tuxedos. We set it to ‘Events’ so that we can pitch multiple accessories, cross-sells and upsells like ties, wallets, shoes for all events and even car supplies like waxes – everything needed to make an impression at posh events.

    A couple of other examples are:

    • E-commerce: If a visitor searches for an iPhone, we can set the interest to Smart Devices and start pitching touch pods and iPads.
    • Content & News: If a visitor searches for mostly sports updates from Manchester United, we set the interest to British Sports and suggest relevant stories even from gossip and economic sections if they have to do with Man U and UK sports.

    As discussed, interests can also be used to take the recommendations off-site. If a visitor to a bank is particularly interested in credit cards, he can be emailed credit related blogs and then asked to enroll into a premium credit program.

    How Does Personyze Use Interests?

    Interests are used to update the ‘Interests’ field on Personyze user profiles and then are accounted for by the recommendation algorithm to suggest products, content, and more.

    How Can Interests be Set for Products?

    Interests are set in a number of different ways:

    • Manually: When you upload product catalogs and article lists, you can associate each entry with interests from your understanding of your visitor, or you can go to manage interests, and click add products to interest.

    Screen Shot 2017-04-25 at 7.57.01 PM

    You can also send JS code containing the interest name, or set a campaign audience that will send the interest event, as alternative methods,  use visitor CRM data; really, just about any kind of visitor behavior can be used to determine ‘Interests’.

    Screen Shot 2017-04-25 at 8.05.25 PM


    How many Interests can a Visitor be Tagged with and Can the Interests be Prioritized?

    One Personyze visitor profile can be associated with up to 15 ‘Interests’ and you can:

    • Set Priority for These Interests

    • Indicate Which Interest is the Most Recent

    So, if a visitor is tagged with the interests “iPhone accessories” and “iPad accessories,” and the latter is the more recent ‘Interest’ as detected by the Personyze system, when he lands on the Apple products page, the recommendations lean heavily towards iPad accessories.


    Set up a Link Between Your Site & Personyze, to Auto-Update Interests

    With this link, your website sends ‘Interest Events’ such as liking a particular product, adding items to cart, downloading PDF of an article, and More. Click Here for details about setting up Personyze to monitor product/content interactions.

    Why Does Personyze Need This Data?

    When Personyze has an up-to-date report of the products viewed by visitors, articles read, items added to cart and wishlist, and so on, it can recommend items that resonate with the LATEST (and thus most relevant) interest of a visitor. This eliminates obsolete suggestions.

    For example if a visitor has shown interest in an iPhone then upon his very next visit, one recommendation widget can display other iPhone models in terms of memory space and color, and another widget can display most-bought iPhone accessories like screen protector, liquid damage insurance, and more. This also applies to content and news sites, and other platforms.

    What Kind of Information Can I Pass on From my Site to Personyze?

    • Products – viewed/added to favorite/added to cart/bought

    • Articles – read/shared/commented/saved for later reading/read in full

    • Interests  - main interest / sub interest 1 / sub interest 2

    As already discussed, you can either pass on Interests that you have decided upon, or keep passing on product and article data to help Personyze learn and evolve in real time.

    How Do I Set up the Link Between My Site & Personyze?

    Whenever a visitor interacts with the elements on a page of your website, Personyze needs to receive a comprehensive report with the ID number and type of interaction in real time. This updates both existing interest data on Personyze profiles and helps initiate new recommendations in the same session itself.

    This report can be generated in two ways:

    a. By setting a Javascript code that runs whenever an interaction takes place. There are different types of code snippets for different types of interactions.

    Screen Shot 2017-04-25 at 8.08.51 PM

    b. If all the data needed for the report is contained on site pages, You can use the Simulator to set containers that identify the sections/places from which Personyze can extract data and use it to update ‘Interests’. Everything from site search data to URL parameters can be extracted by the Personyze system.

    Screen Shot 2017-04-25 at 8.09.16 PM


    What happens when my site reports to Personyze?

    With the reports on user interactions from the site, Personyze updates the user profiles within the system with the latest interest information. This helps the system keep track of latest and most dominant preferences, and if needed, modifies recommendations in real-time, as the session progresses.

    **Customize Recommendation Widget Look, Feel and Design and Choose the Algorithms for Accurate Suggestions.


    Can I Have Different Widgets with Different Logics?

    You can have different widget logics – both on the same page and across the site. As an example you can suggest most visited product categories on the Home page and then on the sub-page of a particular category you can have one widget suggesting accessories, and another displaying most purchased products.


    Can I Customize Recommendations?

    You are in control of your recommendations. You may wish to display certain products or articles to certain visitors based on characteristics like age, marital status, and location. In order to set this preliminary data, you need to add the information to the catalogs and lists you update into the Personyze system.

    For example, if you are a jewelry shop and have many selections of rings, you can tag the engagement rings with relationship status ‘engaged,’ to only suggest them to those visitors who are looking to purchase engagement rings.


    Can I Choose the Look & Feel of the Recommendation Widgets?

    Personyze has many customizable templates for its recommendation widgets. You can have the suggestions go native and embed them to make them look like a part of the site. Or you can have them pop/float it.

     Screen Shot 2017-04-25 at 8.14.36 PM


    Your Site With Recommendations Set:

    When a visitor comes to your site and the Personyze recommendations system is up and running:

    Repeat Visitor:

    • Sees recommendations and suggestions for products, articles, news items and general content which automatically populate the widgets that you have chosen for the pages based on existing data on past interactions.

    • Current session interactions are sent also sent to the Personyze system as reports in real time and they refine his Personyze profile.


    New Visitor:

    • Sees recommendations and suggestions in the pre-defined pop-up or embed templates but these are based on the ‘Interests’ that you had initially tagged each product/item of the catalogue or list with.

    • As he/she starts interacting with your website, Personyze immediately starts learning about him from the reports it receives from the site, and refines the initial recommendations and suggestions.