To make the most effective product recommendations, dotmailer has a data enrichment feature that uses artificial intelligence to find, extract, and classify extra information about your products. This feature analyses your product images and webpages, then extracts and categorises information including descriptive labels, keywords, and colours into Insight data tags.
To use data enrichment, your Insight data collection must include either an
image_path field with a URL to an image of a product, or a
url field with a URL to the webpage for a product. For best results your Insight data collection should include both of these fields.
Types of enrichment
Data enrichment is extracted in the following ways:
|Type of data extracted||Required Insight data field||How it works|
||Artifical intelligence analyses the image and provides detailed descriptive tags, which a user can create filters against in the builder. For example: shirt, short sleeved, summer, collar.|
||Artifical intelligence identifies dominant colors by the hex value. These values are then abstracted to their named colour. This means a user creating a recommendation can create simple descriptive filters like "blue".|
||The meta tags are extracted from the product webpage.|
When you've added the required Insight data fields to your Insight data collection, you can enable product data enrichment to start the data extraction process.
Enabling product data enrichment
To enable product data enrichment on your account, go to Settings > Insight data > and click the Properties icon for your product catalog. Here, you will find an option to enable automated data enrichment.
Once enabled, we will begin to automatically enrich your product data. The initial enrichment may take several hours for very large catalogs. New products that are added to your catalog will be enriched automatically when they are synced to dotmailer.
After data enrichment has completed, your Insight data collection will have an additional field labelled 'data_enrichment', which contains an array of tags that you can use to filter your product recommendations. For example, you may want to filter your best sellers by jackets with a hood.