How Your Data Import Impacts Generation Quality

Wondering how you can achieve top quality outputs from AI? Consider your data!

"Garbage in, garbage out": One Describely customer coined the idea, and we can't disagree. AI can work a lot of magic ✨ but the quality of your data impacts generation quality. Here are our tips for getting your import right!

In this article

High Quality Import Example

  • Clear, descriptive title: Context and detail enables the AI to perform at its best. Make sure your title or existing description is accurate and as thorough as you can manage.
  • Include a SKU or other identifier: If your product data is minimal, that's okay! Including a product identifier in your upload allows Describely's data enrichment feature to pull in product data from across the web for you.
  • Keywords: If your import file has a few or lots of columns for assorted product information, choose to import the most important, value-adding ones as keywords. Describely will take this context and incorporate it into your product copy.
  • Clear dimensions: When you import dimensions, be sure to include the type of measurement in every row OR within the column header. Otherwise, the AI may try to fill in the blank with its own assumptions.

💡 Big takeaway: Treat the AI like you would a colleague seeing your data for the first time. If they can't understand what they're looking at, chances are the AI will struggle too.

Low Quality Import Example

  • Minimal title: Notice the difference between the two images above? One is very general while the other captures specifics that make the product unique and desirable to buyers.
  • Lack of SKU or product identifier: Including a SKU is key when your product data is minimal, as it sets up Describely to pull in product data from across the web via our data enrichment feature.
  • No additional keywords: If your title is descriptive or you imported an existing description, keywords aren't necessary, but if your product data is minimal, importing at least a few descriptive keywords about your product's main value propositions is super helpful in guiding the AI.
  • Confusing dimensions: This one is easy to overlook but essential for AI accuracy! When you import product dimensions, make sure you include the measurement type (inches, centimeters, etc.) in either the columns' header or in every row.
  • Obscure product data: When importing product data, avoid including abbreviations that aren't well-known as they can confuse the AI.

💡 Big takeaway: We understand you are forced to work with minimal product data at times (or maybe ALL the time) and that's okay! Even with just a product SKU or other identifier, our data enrichment tool can help fill in the blanks.