Algolia Recommend and Personalized search require a WPSOLR Enterprise license.
This documentation describe the configuration of your indexes for Algolia Recommend and Personalized search.
If you wish to learn how personalization improves your website and how it works, you could check out our general guide on recommendations.
Algolia recommend and Personalize
Algolia search is already part of WPSOLR Pro. With WPSOLR Enterprise, we unlock some of the incredible features supported by the Algolia Premium plan:
1) NeuralSearch (requires an Algolia Elevate plan)
2) Analytics
3) A/B Testing
4) Dynamic Re-Ranking
5) AI Synonyms
6) Query Suggestions
7) Query Categorization
8) Recommend
9) Personalization
Activate your first-party cookie
Enabling the first-party cookie will generate a unique id to visitors, anonymous or logged in. The AI models will be able to follow this id across the many visits a visitor can make to your site before converting.
- Select the search tab
- Select the search menu
- Select a view, default or custom
- Select your Algolia index
- Select “Use a visitor cookie”
Activate user events
Collecting user events is done automatically for you by WPSOLR Enterprise. No need to rely on tedious external tools integrations like Google Analytics or Google Tags.
- Select “Capture visitor interactions” to send visitor history in real-time with our custom Javascript code
- If necessary, set the CSS selector of your search page results here. This will enable to detect visitors clicking on Algolia search results even as results are displayed by your theme. Notice that here you should pick a CSS for all Algolia produced archives: search results, shop results, categories, tags, authors, year, month, day, post types.
- Choose if click events should be sent several times if not voluntarily triggered
- During the setup phase, you can show events sent in the Javascript console of your navigator
Send historical user events
As mentioned in the general presentation of recommendations, sending user events is crucial to get increase your current conversion rates with Algolia personalized search and Algolia recommend.
The first step id to configure your backend to send historical events, like purchase orders and rating stars. Fortunately, it is easy with WPSOLR Enterprise.
- Select search tab
- Select data menu
- Select your previously created Algolia index
- Select your events in the event section. For instance the WooCommerce legacy order events, or the new HPOS orders
- Select the status of the events to be sent
- Also select the post types that will be later selected for display in your recommendations
Train AI models
Algolia Recommend provides four AI models to build recommendations upon.
Before using a model within WPSOLR to display recommendations, you need to:
- Login to Algolia
- Select the application where you created your Algolia index
- Click on the “Recommend” menu (left menu with a red bulb)
- Click on the “Models” menu
- Click on one AI model
- Configure ans train the model:
1. Select the Algolia index configured in WPSOLR and indexed with your catalog and events
2. Algolia will warn you if you have enough events collected to start the model training
3. Deduplicate recommendations
4. Start the training
- Wait a few hours to few days until the model is marked as trained
Below is a screenshot of 3 models trained, one failed due to a lack of events.
Create WPSOLR Recommendations
- Click the “Search” tab
- Click the “Recommendations” tab
- Select a view configured with an Algolia index
- After saving your recommendations, you can pase this shortcode anywhere in your content: home page, product detail page, blog post page…
- Choose a model among the 4 models provided by Algolia. The model needs to have been configured and trained first in your Algolia’s account dasboard
- Choose a layout to display your recommendations as you wish. You can also build your own layout, or hire us to build it for you
- Number of recommendations displayed
- Show the description or not
- Size of the featured image. May not be useful, as most of the layouts take car of the image size for you
- Use custom CSS to tweak the layouts with minor modifications like border colors, font sizes, text colors…
- Select the post types you want to show in recommendations. They must have been selected in screen 2.2 and indexed
- select the WooCommerce features to show: star rating, product price, add to cart button
Notice that you can repeat this operations to add as many recommendations as you need.
Check events sent to Algolia
During the setup phase, you should verify that events are sent correctly. Log in your Algolia application, and navigate to the events dashboard as shown below.
Setup personalized search
Algolia personalization is based on events. You select your events on a strategy, with a score.
For instance, do not select add to basket conversion events, or select purchase order events with a high score, to boost orders and discard add to basket.
Analytics dashboard
Collecting events will unlock advanced Algolia dashboards, as shown below.
You can see statistics on clicks, empty results, and conversions.