Period: Jul/Aug 2023

Tweakwise App & API

• We improved the performance of the Derived Attributes Combine module (in particular, the attribute select boxes).
• We fixed a bug in the language switcher where it would always show the English flag.
• A small bug with the Tweakwise logo not being responsive was solved.
• Apparently, it was not possible to save an autocomplete configuration if the selected category had ID 0. It would instead give an error that no category had been selected. That has now been solved.
• The next frontend API endpoints now also support the ‘tn_lang’ parameter:
  • suggestions/grouped/{instancekey}
  • suggestions/{instancekey}
• A bug was fixed where the external attributes preview didn't work properly for Page View reports in the GA4 module.
• We've fixed a bug that caused accrued filter values sorting (order of derived attributes) to not work correctly in some scenarios. For example, the accrued sorting order didn't work at all when used as alternative sorting order (sorting order used when the filter is expanded). Accrued sorting also wasn't applied when used in combination with smart filters; it could be configured to act as a fallback but wouldn't work if configured in that way.
• We released a simplified way for users can switch between their instances once they log in to the app. Instead of being redirected to a separate page with a single select field, they are now presented with an in-app context menu with a list of their instances and will be redirected to it after a single click. At first glance, visually, everything stays the same.
• We fixed a bug where an up- & cross-sell recommendation with configured parameters could not be deleted.
• It was possible for a publication task to fail if the instance used a derived attribute that took a lot of time to generate the values. We’ve effectively increased the amount of time this process is allowed to take, preventing the publication from failing because of it.
• We’ve added a modern look to our AI synonym suggestions, so it will be instantly clear that AI is involved in certain modules. This look and feel will come back in further AI implementations.
• Google Analytics4 import would fail on data from Google being provided in an unsupported format (Scientific notation of numbers: 1.04e-05). This is now fixed, so GA4 imports should not fail on this anymore.

Javascript

• We've made some changes for a smoother integration. Developers implementing the JS Implementation can now programmatically open and close the search results overlay. Also, we've fixed a bug when programmatically "refreshing" the JS Implementation.
• We added a class to the pagination component when it only has one page, which can be useful if you want to hide the complete pagination in case it makes no sense.
• We've added a grid/list view functionality for the displaying of products. This feature is optional for JS implementations and can be implemented in different ways. We can show only a grid, only a list (new), or a version with both grid and list view (new). For the last one, we implemented a switch functionality too. The choice made by customers will be recognized for reoccurring visits to the webshop.
• We have made a few small changes to JS implementation filters (adding summary for slider type & configurable threshold for hiding top filter facets) and added metadata to the starter.js file:
• Fixed a bug where the (price) slider could break when first using the slider and then using a regular filter.
• The "input" fields (which were needed for technical implementation) are now optional instead of required. As it shouldn't be required in combination with the Suggestions JS.
• An option was added to the Guided Selling JS that non-js-implementation customers could use to redirect to an existing (non-js) lister page. Please note that custom JavaScript code/logic is required to correctly pass on the selected filters to the (non-js) lister page.
• Sometimes when selecting a single filter, the selected filter summary didn’t show up; this has now been solved.

Magento

• When a product belongs to multiple categories, and there are some recommendations configured for one of those categories, we now show the right recommendations. In the previous versions, the category recommendations only checked the first category the product belonged to. We added a fallback category for recommendations as well, so we can always show something relevant.