Migrating from HTTP to HTTPS URLs is generally easy to do. However, we offer the below advice and best practices to ensure we are able to successfully identify duplicates and to help you manage your inventory.


For JavaScript integrations (automatic inventory scraping):

If you are not updating the canonical URLs from HTTP to HTTPS you can enable HTTPS for your site at any time without affecting our recommendations. We use the canonical URL as the unique identifier, so all inventory items and activity tracking will continue to use HTTP for as long as the canonical URL is also HTTP.


If you change the canonical URL we may temporarily have duplicate versions of items within the inventory (one HTTP and one HTTPS). Our deduplication logic considers both URL and title - as long as the HTTP and HTTPS versions have the same title you will not see both URLs recommended in the same page. If you have performed a hard cut-over and there is no way to access HTTP URLs after the canonical URLs have been updated all old URLs should expire 30 days after they were last collected. 


If both http and https versions can be collected (i.e. the canonical depends on which protocol is used) this can cause duplication issues if only one version is updated with new metadata, or if only one version has the title changed. While these issues are uncommon, they can lead to duplicate recommendations appearing if there is a mismatch between both URL and title. We can implement filters on both allowed inventory and allowed recommendations to get and return only one version. However, activity tracking in a mixed-protocol environment can affect performance as the data is essentially split between each version of the URL. We strongly recommend setting canonicals to explicitly use one protocol or the other, and to minimize the usage of mixed-protocol environments.


Remember:

  • Make the switch all at once if possible - mixed protocols can prevent accurate activity tracking
  • The ID used is tied to the canonical URL - nothing will happen until the canonical URL is changed
  • Most inventory items will expire 30 days after they are last seen by our JS - avoid title changes for one month after switching to ensure our deduplication can catch duplicate items based on title.
  • If duplicate items begin appearing please notify Support - we can block the previous versions of the URLs from being shown in the recommendations, and migrate activity data to the new equivalent URL to minimize performance loss.


For API (using URL as the unique ID):

Please notify Support so that we can assist you in a successful transition! We generally recommend adding all items with their new HTTPS URLs, then allowing us to migrate the activity data from the old versions of the URLs to the new ones before deleting the old HTTP URLs. This process can take several days to complete, so please notify us ahead of time so we can coordinate with your team. There may be a slight performance drop during the transition.


For API (using ID):

If your inventory uses ID rather than URL as the unique identifier, you can simply update the item IDs with the new URLs and their corresponding metadata. There will be no affect on performance because activity is tied to the ID of the item rather than the URL.