🗞️ Use AI to be TOS compliant by August 15th 💥💪

Welcome to the 6th edition of our new AI for E-Commerce newsletter, your trusted resource for actionable AI strategies in eCommerce, especially on Amazon! We have now crossed 1000 subscribers💥!

In each issue, we spotlight practical AI use cases, showcase our favorite cool tools that you can start using today to improve your eCommerce business. Our focus is on tangible benefits. No technical jargon, no pie-in-the-sky concepts—just straightforward, actionable advice to help you stay competitive and grow your online business. 

Let's go...

For this first section on AI Practical Use Cases, we’ll talk about how to use AI to clean up the house before the auditors show up (literally!) Some of you might have already seen this, but Amazon has given us until August 15th to fix our listings or face consequences! Read this post by our friend Robyn Johnson on LinkedIn.

So, how can you use AI to clean things up, quick. Let’s test out ours strategy on this Turmeric supplement that is using All-Caps in their bullet points, which is a no-no.

I saved the text of Amazon’s guidelines into a PDF and uploaded this document directly into ChatGPT. You can get a copy of my PDF here.

I then asked it to fix the text of my bullet points without entirely rewriting or dropping important keywords in this prompt:

“Based on these guidelines, correct my bullet points, WITHOUT dropping any important keywords or changing my wording or style. At the end, tell me what you have removed.”

Just for fun, I had intentionally inserted a phrase related to refunds that I know will get flagged by Amazon.

As you can see, it did a fantastic job, removing just the problematic parts where “refunds” are mentioned, and correctly changing upper case to lower case throughout. It also correctly told me what it had changed and why.

If you have thousands of listings that need to be fixed in built, consider using a ChatGPT or Gemini AI integrated Google Sheets plug-in. There are several tools out there on this market, and here is one.

Also, I tried to convert my method into a CustomGPT but it appears that since the training is happening on protected Amazon data, OpenAI rejected my GPT for Public use (I can still use it personally), but I lodged an appeal which is currently under review. Stay tuned!

For the Cool Tools section today, we are talking about RunwayML. This has a cool feature of converting any still image into an animation💥!

So here is one image that I borrowed from an Amazon Post from NatureWise. I kind of want to give the girl’s hair a little movement.

I use the Motion Brush tool to mark sections of her hair to animate.

And here’s the outcome. Pretty impressive!

Here I used RunwayML’s Gen 2 model for free. To use the latest Gen3 model (which we here is INSANE), you need to upgrade, How much you ask? It’s $12 per editor/month! Check it out!

For this week’s Nerd Bytes section, I’d like to touch on Vector Databases. Last week we talked about RAG (Catch last week’s newsletter here). This week we go further and explore some important concepts such as:

• Unstructured Data: Data that doesn't follow a predefined format, like text, images, and videos.

• Nearest Neighbors: A technique to find the most similar data points to a given point in a dataset.

• Clustering: Grouping similar data points together based on their characteristics.

• Word Vectorization: Converting words into numerical vectors that represent their meanings and relationships.

• Indexes: Data structures that improve the speed of data retrieval operations.

Essentially, all types of data such as text, audio, video etc. need to be converted to numbers so that computers and LLMs can understand them.

I found a really easy-to-understand video (~4 mins) that talks about these concepts:

Building a vector database requires programming skills, but there are also some ready to use solutions and pre-trained models that can help you get ahead if you don’t want to go the programming route.

Some practical use cases of Vector Databases:

• Convert a body of knowledge such as your social listening data into a vector database that can then be queried with LLMs

• Convert organizational training from videos, documents, slides etc into a vector database that can be used for queried with LLMs

• Assessing the semantic similarity of products. This is similar to what Amazon’s RUFUS and COSMO are doing already.

While most of us may not have the skills and knowledge to create vector embeddings on our own, it is important to watch this evolving space and be aware of how it might impact you as an Amazon seller.

For the Thought leader section this week we asked our friend Matt Kostan from ProductPinion about the future of AI in E-Commerce and this is what he had to say:

“The way people search is changing.

Search is becoming more conversational and is now happening on AI platforms like Rufus, ChatGPT, and Perplexity.

If you’ve ever tried Perplexity, you’ve seen how the experience is ten times better than Google. OpenAI just announced their SearchGPT prototype last week, directly challenging Google and Perplexity. The shift in search is here.

As a Canadian who plays hockey, there's a saying: you need to go where the puck is going, not where it is.

So, how can sellers ensure their brands and products are recommended by AI platforms?

The first step is to list your products on reputable sources. For product brands, it’s easier than you think. Major publications have product roundup sites that actively look for products to list. There's New York Times Wirecutter, Forbes Vetted, USA Today Top 10, CNN Underscored, etc. Reach out to the authors and pitch them on why you should be listed too.

Search platforms use these sources to answer queries about preferred brands. As a side benefit, watch your listing conversions skyrocket once you add a trust badge like “As seen in Forbes.”

As a bonus tip to be ready for conversational search, run your listing through a Pinion Ask on www.productpinion.com and ask real shoppers how they would describe your product.

You’re guaranteed to find semantic keywords your listing is missing (I was blown away). Plus, since you're asking real shoppers, you'll get very specific results instead of the generic answers typical AI research sometimes gives.

If you’ve loved this and want to hear more conversion optimization strategies, sign up for the Conversion Lab Newsletter. Once a week, you’ll get a specific strategy that helps you make more sales. For more on the tips mentioned above, feel free to email me at [email protected].”