Effective Product Attribution Q&A with Lily AI’s Sean Gouldson

Sean Gouldson, Lily AI’s Head of Presales, discusses some of the top questions asked about Lily AI’s product attribution management capabilities.

Granular product attributes can make all the difference for retailers and brands when their products are attributed comprehensively in the language of the consumer, as opposed to falling back on legacy, generic out-of-the-box attributes.

But how exactly does this process of attributing products more effectively work in order to drive bigger cart sizes, stronger conversation rates, and more? We sat down and talked with Sean Gouldson, Lily AI’s Head of Presales, to discuss some of the top questions retailers and brands want to know about the ins and outs of product attribution data.   

Sean Gouldson

How is product attribution applied? What are the nuts and bolts of how this works? What does the item set-up with Lily AI data look like?

When it comes to a retailer’s process for onboarding a new product and getting it ready to sell digitally or online, there is a lot more to it than meets the eye. There’s so much that comes into play with this. For instance, how a retailer may index for search, what page the new product is going to show up on, how filters and facets will be deployed for navigation, and more. Product attribution data can help to tackle all of these behind-the-scenes internal processes with an optimized item set-up or better onboarding practices. 

So, what does this look like with Lily AI’s help? Lily AI works to take the retailer’s current data to either improve or replace some of the existing processes or workflows. For example, a retailer is already going to have assets present – they’re going to have an image and text – but this still may not be enough to offer something for sale online. 

The next steps include passing this present data (the images and text) to Lily AI in batches where we will give the retailer the most accurate and relevant product attributes back. 

Lily AI takes the data feed and breaks it into batches of similar products. After this, we will:

  • Run the data through two different AI-powered processes:
  • Run the text through text recognition and Natural Language Processing (NLP) pipelines. 
  • Combine and leverage the image and text together (image and text based predictions together get better results than separately). 
  • Loop in the human element of our domain experts team who may review a selection of the data for better accuracy.    

How Lily AI Works

Does the foundational product data coming from a retailer or brand need to already be strong to gain the benefits of Lily AI enrichment and enhancement?

The foundational product data coming from a retailer or brand doesn’t already need to be strong in order to gain the benefits of Lily AI’s enrichment and enhancement capabilities. As long as you have images and basic text that could be used for non-visible details (i.e., materials, lining, etc.) your product data is viable, and Lily AI can take that as the input.  

Does Lily AI pull all of the product attributes from historic images? What does a retailer or brand need to provide Lily AI?

Lily AI pulls as many attributes as we can from the images and text that is provided for us. Historical products refer to products that aren’t new, but are currently on sale. The images we pull product attributes from do not have to be historic. We like to keep it simple and as long as we have those images and text we can combine them, improve upon them, and send the retailer back a single set of data.  

Do subjective tags change over time (based on season, for instance)?

Generally the tags don’t change over time, because the merchandise is still the same. Although the merchandise may be intended for a specific season or occasion, the only way Lily AI would alter the tags is by offering additional ways to describe the merchandise as we see trends change. The way a consumer may describe products with their own language and their own concepts may change, so Lily AI may add things over time to keep the tags rich with that updated relevant data. 

The additional ways to describe merchandise due to seasons or emerging trends may very much evolve, therefore Lily AI can add things to keep the tags as fluid as possible. 

Do Lily AI teams review tags to ensure consistency with the brand, or is it all automated?

This depends on the customer and their own specific needs. Lily AI will work to find the right approach, while our teams of experts can review the automation and its results as necessary. 

Can the AI read lifestyle images, or just product imagery? 

A man in fashionable clothing

Lily AI can do both! We can read lifestyle, multi-product images, high quality catalog photography with photos of apparel (e.g., mannequins, hangers, flatlays), and/or photos of clothing models.   

See Our Product Attribution in Action

Trusted by global retailers, brands, and industry leaders, Lily AI can help you to connect your customers with what they’re really looking for. See what we can do for you today with our product attribution data capabilities by setting up a time to chat.
Request a Demo
Woman in a yellow sweater smiling at her laptop while shopping for clothing.