Overview
The WELCOME WPO is about offering personalized product recommendations on your start pages (e.g.: home page).
If you are not familiar with WPOs, Customer Journey Steps, WPO Levels, Widgets, or Page layouts, please read the introduction: Widget & Page Optimizers (WPO)
Purpose
Optimization of product recommendations on the start/landing pages to start the visit as relevant as possible and reduce bounce rates.
Optimal alignment with the origin of the visit, prior knowledge of the visitor, and your current business goals.
How should the widgets appear visually?
The WELCOME widgets typically are displayed with a label (title) on top and a slider or product recommendations (with left and right arrows) as documented as “product slider” in the https://boxalino.atlassian.net/wiki/spaces/BPKB/pages/57671688/Check-list+empower+your+Website+Layout+with+Narratives#Layout-Blocks-(Templates)-Check-list. There might be several blocks on the page, one below each other, or in different sections of the page.
Key focus: personalization
Personalization is important here because your start pages have a very broad context (on your home page, every product of your e-shop might be considered, while on a search or category page, only the product matching the search or the category are in context). Therefore, instead of “simply” showing your global top-sellers, current promotions, or novelties, you can instead (or in addition) select products matching the prior click and purchase behavior of each visitor individually.
Related WPOs
The WELCOME WPO belongs to the AWARE customer journey step (first column of the diagram above) and corresponds to LEVEL 1 as it is solely about products and not other types of content: combine it with PROMOTE, READ to also support recommendations of other types of content like banners or blog articles and use it within a smart, personalized and dynamic layout with PRESENT.
What widgets should you consider?
One or several product recommendations blocks, each with a specific label, can be integrated into each start page.
Each block can display the products in a slider or in a grid (the number of products per block is not limited but is typically less than 20). The number of blocks is also not limited and will be returned in a sequence by our API. Combine it with PRESENT to retrieve a dynamic personalized list of blocks embedded in a complete page layout (including other visual blocks for banners, information messages, emotional pictures, … all of which can be also personalized and A/B tested). As an example of such a case, here is a use case with Marketing Topics: Personalized Marketing Automation on Qualipet Homepage
WPO Optimization Strategies
The Widget strategy can be configured in the Widget Strategy Editor and supports all the standard Strategy Use-Cases.
Here is a selection from our Best Practice Strategies in 3 sections (what to configure before your go-live, what could be your first A/B test about and more advanced practices for later stages).
1. What to start with (for the go-live)?
Group | Name | Short description | |
|---|---|---|---|
| 1 | Best-selling trends, top recently generating revenue / margin products, often contributing to higher AOV, … | ||
| 2 | Exclude Ineffective Product Groups | Remove out-of-stock products, as well as zero or low-value products, low margin, or any other unwanted products (the second practice considers the 1st or 2nd most sold sku of each product group, for example to exclude products which top sold variant(s) are currently out of stock) | |
| 3 | Promote products or content with valuable value flags (5 stars, best-seller, novelties, recent discounts, staff pick of the week, …) | ||
| 4 | Promote products or content with attribute values matching previously viewed products | ||
| 5 | Promote products or content with attribute values matching previously bought products | ||
| 6 | Exclude products the customer has recently purchased | ||
| 7 | Exclude Products already in the basket | Exclude products that are already in the basket of the visitor |
2. What could be a/b tested first?
All the elements of the prior list (What to start with) could also be used in a/b testing. The list here-after provides additional Best Practices that can be quickly tried.
Group | Name | Short description | |
|---|---|---|---|
| 1 | Promote products or content with Discovery Tags matching previously viewed products. Each product gets a list of Discovery Tags representing the most common search terms used to find a product. | ||
| 2 | Promote products or content with Discovery Tags matching previously bought products. Each product gets a list of Discovery Tags representing the most common search terms used to find a product. | ||
| 3 | Previous Search(es) Related Search(es) | Promote products matching the prior searches (or related to the prior searches) of the visitor. | |
| 4 | Promote products or content in the target of active promoted campaigns | ||
| 5 | Neural Collaborative Filtering for individual recommendations based on the purchases of the customer and leveraging the patterns in the purchase behaviors of other customers | ||
| 6 | Basket > Often Bought Together Basket > Collaborative Filtering Bought Together | Promote products often bought together with products already in the basket of the visitor (first practice with statistics, second variant with AI Collaborative Filtering) | |
| 7 | Views > Often Bought Together Views > Collaborative Filtering Bought Together | Promote products often bought together with products the visitor has already viewed (first practice with statistics, second variant with AI Collaborative Filtering) | |
| 8 | Last Purchase > Often Bought After Last Purchase > Collaborative Filtering Bought After | Promote products often bought after the last purchase of the customer (first practice with statistics, second variant with AI Collaborative Filtering) | |
| 9 | Purchases > Often Bought By the Same Customer Purchases > Collaborative Filtering Bought By the Same Customer | Promote products often bought by the same customer (based on the purchase history of the customer) (first practice with statistics, second variant with AI Collaborative Filtering) | |
| 10 | Basket > Often Basketed Together Basket > Collaborative Filtering Basketed Together | Promote products often basketed together with products already in the basket of the visitor (first practice with statistics, second variant with AI Collaborative Filtering) | |
| 11 | Views > Often Viewed in Same Session Views > Collaborative Filtering Viewed Same Session | Promote products often viewed in the same session as products the visitor has already viewed (first practice with statistics, second variant with AI Collaborative Filtering) | |
| 12 | Purchases > Repurchaseable products | Promote products the customer has already bought which can be repurchased Each product can be automatically labelled as repurchaseable if enough people bought it several time, or if it belongs to a category you know to be repurchasable or doesn’t belong to any category which you know to be not-repurchasable. |
3. What else could be experimented with?
Group | Name | Short description | |
|---|---|---|---|
| 1 | Promote products or content in the same Price Quadrant as previously viewed products. | ||
| 2 | Promote products or content in the same Price Quadrant as previously bought products. | ||
| 3 | Wish-list > Products Wish-list > Products Attributes Wish-list > Products Discovery Tags | Promote products
| |
| 4 | Basket > Products Attributes Basket > Products Discovery Tags | Promote products
| |
| 5 | Best-selling trends within the Cluster the visitor is predicted to belong based on his prior product views. Customer Clustering distributes all customers into clusters based on their purchase history | ||
| 6 | Best-selling trends within the Cluster the customer belongs. Customer Clustering distributes all customers into clusters based on their purchase history |
Additional ideas for Widget Strategies
Request 4-5 highly specific blocks and only display the first 2-3 returning some recommendations
Change the label of your blocks depending on the content of the recommendations (e.g.: “Your personal suggestions” versus “Our best-selling articles”)
Segmented personalization (based on the visitor’s traffic source, device, etc.)
If your widget is labeled “bestseller”, consider setting up a FILTER with a minimum number of sales or with a bestseller label in the PCA - Value Flags Best Practice to avoid the risk that personalization starts showing products with lower sales levels
Onboarding project plan
Here are our suggestions for the project management steps of the Welcome widgets.
This process described how to integrate the WELCOME WPO by itself without any other WPO combined. if you want to integrate the WELCOME together with other WPOs, contact Boxalino to discuss an adequate project plan.
Task | Description | Comment | |
|---|---|---|---|
| 1 | Pre-requisites | Make sure the Pre-requisites steps are completed and already deployed in production | Exports for products and transactions are required, as well as all the standard tracking events. |
| 2 | Pages | Define the pages where the WELCOME widgets will be integrated | Example:
|
| 3 | Widgets | Define the WELCOME widgets to be integrated on each page and where they should appear on the page (if you decide that some widgets should not appear for all visitors, define the logic deciding when they will appear or not) | Example names:
|
| 4 | Labels | Define the label of each of the WELCOME widgets in each language (the title that appears on top of the product slider) | Examples:
|
| 5 | Integrate API + Configuration | Integrate Boxalino Narrative API and configure your widgets in the Boxalino Admin.
| As a result, you should be able to see the widgets appear in your dev/stage front-end showing valid products but without any relevant logic in the selection of the products. |
| 6 | Configure Widget Strategies | Define the strategy of each of the WELCOME widgets (first in prose and then by configuring the widget in the Boxalino Admin) | Boxalino can support you for the configuration of the widget strategies |
| 7 | Test & Deploy | Testing in your dev/stage environment and go-live | In case you have a separate prod and a stage account, make sure to copy your configuration in production before going live |
Integration notes
Pre-requisites: Data exports (products, transactions, and optionally customer data) and Boxalino Tracker integration.
Boxalino Narrative should be configured in Boxalino’s admin on the first appearing widget on the page (which should be the one called in the API request). The Narrative should define the layout of all the widgets (name of the widget to be indicated in the accessor parameter possibly with a hitCount to define how many products should be returned, for example: “topsellers[hitCount=15]”) as well as the labels.
Boxalino Narrative API will return a list of blocks (for each block of product recommendations) each with a list of sub-blocks (with the list of products) as documented here, (make sure to make only one call to our API to retrieve all widgets and not a call per widget which would cause poorer speed performance on your frontend)
The products returned will always include the product ids (which might be all you need) and other product fields can be returned as well if requested in the API calls (see Return Fields)
The request and the response should not be cached (consider an AJAX call in case you need to cache some parts of your page for performance reasons).
Make sure to tag your HTML with the required classes and entities returned by Boxalino as bx-attributes so the Boxalino tracker can identify automatically the scrolling behavior (including the appearance of the products when using the slider arrows).
Before going live, make sure to control the Tracker Checklist
Add Comment