How to gain competitive edge by using generative artificial intelligence (Gen AI) in online marketplaces

Online marketplaces are a maze for both shoppers and business owners. No matter how much you invest into building a powerful, frictionless machine that sells while you’re asleep, there’s always something else to do and improve.

According to Stanford’s Persuasive Technology Lab, almost 75% of people judge the credibility of your website based on ease of use and design. But that depends on the current standards, right? 

If the checkout process is complex and slow, 17% of shoppers will drop out. But what if you want to increase cross sell purchases and lower product returns? Or follow your mission towards sustainable and transparent business, which entails some friction?

Complexity doesn’t stop there. For example, the highest cart abandonment rate is in the APAC region. Yes, the data differs depending on the region, economic, social, and other factors. But how can you predict that on time?

We’ve got heaps of data about what customers want. We have research and statistics. Shoppers tell us everything about their habits every day through clicks, feedback, ratings… 

But the real challenge is making sense of this heap of information and putting it to use – in real time – to make customer experience better. 

So far, we’ve been relying just on human analytical capabilities, combined with various tools. There are plenty of tools out there that can crunch the numbers, but the one that promises to change everything is Gen Ai. 

Read on to find out how Generative Artificial Intelligence can help your online marketplace flourish.

Types of Generative Artificial Intelligence tools that are ideal for e-commerce

Data without interpretation is just noise. The biggest power of Gen Ai lies in their ability to mimic human creation and analyze swaths of data at lightning speed. 

It’s about making shopping feel less like guesswork and more like a chat with a buddy who just gets it. It’s about giving the customers what they want, when they want it. In the end, it’s combining conversation and search into one experience – making the communication between the interface and people more natural.

If you type in these terms into Google, you’ll find the following categories of tools in the world of e-commerce:

1. Text generation tools that are based on Large Language Models (LLMs). These can craft product descriptions, answer customer queries (chatbots), and even create marketing content.

2. Image generation tools that can produce high-quality visuals, useful for showcasing products that are yet to be physically manufactured or for visualizing customer-specific customization options.

3. Predictive analytics that analyze user behavior to predict future purchase patterns, e.g. inventory management and targeted marketing campaigns.

4. Voice assistants that can help users navigate the platform, make purchases, or get product recommendations.

5. Multimodal systems that combine text, image, and voice. These tools offer a comprehensive AI solution, enhancing user experience by interacting in multiple forms of data.

6. Anti-fraud systems that analyze user behavior and spot malicious intent.

7. Recommendation systems that deeply understand what the user wants and in what context.

Now, it might be tempting to opt in for a solution quickly, even though the list sounds futuristic and almighty, but it’s vital to deeply assess your internal needs.

For example, a marketplace that sells custom artwork might greatly benefit from image generation tools, enabling users to see a visual representation of their custom requests.

But a platform selling books or literature might find more value in text generation tools or voice assistants.

The overarching principle? Assess the specific demands and characteristics of your online marketplace. It’s not the tools that come with a hefty price tag, but their implementation.

How online marketplaces can use Gen AI to grow

Marketplaces are online places where people search for items, right? This technology changes the way you interact with items… And the way search works in general. 

In a nutshell, on a marketplace, Gen AI either affects you or it affects items (or both at the same time). By affecting you, it means that it assists you in making choices. It gives out recommendations and joins you in finding solutions. When it affects items, it helps shape them the way you want them to – changing colors, or categorizing them…

Here are some AI-based tools that affect you and your overall experience:

• Chatbots are the first thing that comes to mind. These AI-driven virtual assistants can handle customer queries, process orders, and even assist in product recommendations. 

Imagine a user asking, “Do you have summer dresses in blue?” The chatbot promptly displays a curated list of blue summer dresses available on the platform.

• Fraud detection and prevention is a necessity because every 39 seconds there’s a hacker attack somewhere in the world. Just as Gen AI gives you the ability to do magnificent stuff, it also enables people with bad intentions. AI algorithms can identify unusual patterns and flag suspicious transactions, adding an extra layer of security.
  

Let’s imagine a scenario where a buyer suddenly orders 50 high-value items and chooses expedited shipping to a new address. The system flags this as a potential fraud and alerts the admin.

2. CYBER SECURITY

• Multimodal search (searching with multiple media types) is something that Gen AI brought into the business space. By combining text, image, and voice, users can search products in multiple ways.

For example, a user uploads a picture of a vintage lamp and says, “Find me something similar.” The system then displays similar products available.

And here’s how AI affects items:

• Personalization is a big thing in online sales. According to research by Google, 61% of people expect that brands optimize the experience based on their preferences. Now remember all that data you gather? Gen AI can analyze user behavior, preferences, and purchase history, location… And provide a hyper personalized user journey.

Let’s imagine a person named John that often buys hiking gear and would like to see himself in a specific color. Or let’s say he’s buying clothes for his girlfriend.

Source: Google Virtual Try-on

Gen Ai could generate models in different sizes, items in different colors and allow John to imagine and see if it fits him… Or somebody else.

• Advanced semantic search is an advancement of traditional search. In traditional search, people type in the keyword and the algorithm behind looks for the exact keyword match. So, if you look for “red sneakers chicago”, if there’s no article with that name in the database, it won’t appear. Or it might show all sneakers just because you typed in “sneakers”.
  
  AI models today understand the context and intent behind a search query.

If you type in “sneakers suitable for marathons”, the system won’t display all sneakers. It will curate a list of shoes specifically designed for long-distance running, while taking your preferences into account if you have already browsed certain colors or sizes.

These applications, when effectively integrated, can drastically enhance the user experience, making shopping seamless, personalized, and efficient. Imagine everything at once.

Imagine searching by text and picture (or voice!) through a chatbot and getting hyperpersonalized results. For example;

“Hey BRAND X, I’d like to buy a pair of shoes. But last time I bought these, and they were uncomfortable. I like the model though. Do you have anything else that’s similar? And please, they need to match my blue jacket. ” and then you attach the pictures.

The chatbot returns the products… And style recommendations. Five minutes later and you’re already rethinking your outfit for tonight.

You probably like this by now, so how can you implement Gen AI?

How to implement Gen AI into your marketplace

As we’ve said, it’s not the tool that comes with a hefty price tag, but rather the implementation and the usage. Understanding the technical aspects of implementation is essential for budget control. 

You have 3 options.

• You can get off-the-shelf solutions that integrate well with your platform and with the Gen AI tools. This is a budget friendly option and the quickest one to implement. We recommend testing out Vantage Discovery, our exclusive partner in conversational search. It works out of the box.

The platform serves as a connection point between a generative model and your platform, which can be a key differentiator, especially for smaller marketplaces.

Another option is the complete opposite of the previous one. You could use your own data to train your models and build algorithms. This is the ideal option if you care about data privacy. You control the resources. Everything. But the amount of resources (infrastructure, data sets, training, specialists…) is way beyond what anyone outside the big league players could afford (think Amazon) and see any ROI in the foreseeable future.

• Another option is the hybrid solution. This means that you build a custom solution on your side that connects with the Gen AI powerhouses. This solution executes the tasks you need – and relies on the computing power of Gen AI.

Example – building a chatbot that’s connected to the chatGPT and that uses your database (the so-called vector database) to provide precise answers in multiple media forms.

Of course, the hybrid solution could also entail integrating third party software for less user-intensive tasks. For example, you might implement chatbots from a vendor, but build a custom recommendation engine, or vice versa.

Limitations of generative artificial intelligence that affect online retail

While Gen AI offers significant advantages, it’s crucial to be aware of its limitations and potential landmines that might slow down your business if not checked:

1. AI can sometimes make decisions that don’t align with human values or societal norms. For example, it might display biased answers based on flawed data. After all, these models are trained on available inputs on the internet, and we know the internet is not a shiny place.

2. Gen AI, especially in text and image generation, can occasionally produce outputs that are inaccurate or don’t make sense. We call these hallucinations. These models don’t think about accuracy. Their goal is to provide an answer.

3. Think of infrastructure from the beginning. As your marketplace grows, the AI system should be able to handle increased data and user interactions without a drop in performance. Some off-the-shelf solutions might struggle with scalability. In fact, many companies think of scalability only later, when it’s already a bit late and great financial damage occurs.

Of course, there are many solutions to at least minimize the influence of the limitations above. Here’s a checklist that we provide to all our potential clients at SmartCat:

• Implement a feedback system. Allow users to flag incorrect or unsatisfactory AI outputs.

• Invest in the infrastructure, especially as user interactions and data volume grow. This might involve investing in better servers or cloud solutions. We suggest taking a look at our partner Pinecone, a vector database that grows (or shrinks) based on your usage.

• Prepare the data. AI operates on data. The cleaner and more structured your data, the better the AI will perform. You might want to take a look at one of our other partners, Databricks. It’s a platform that unifies the data and allows for easy access and processing.

• Keep your AI experts close, either inhouse or as external vendors. Their technical expertise can guide the process, ensuring optimal setup and performance. At SmartCat, we have a Data Consultancy service to help you onboard quickly and assess your AI requirements.

• Test on a smaller scale. This is vital. So many business owners jump headfirst into new tech ventures. Try just one feature before you launch an entire project.

Understanding these limitations and their solutions ensures that you use Gen AI effectively while minimizing potential pitfalls.

Before you commit, why don’t you test out your first Gen AI project?

Gen AI is reshaping the landscape of online marketplaces, offering tools that enhance user experience and streamline operations. 

While its benefits are huge, it’s essential to approach its implementation with a clear understanding of available options and potential limitations. We know this is a true challenge. Especially in times of limited budgets.

That’s why we offer fixed-fee pilot projects to online retailers. Go through our workshop, define a crucial AI-based feature for your marketplace and test it out with your users without breaking the bank (or the entire platform).