The Idea in 60 Seconds: Revolutionizing Online Shopping with AI

  • I’ve written before in this blog about the importance of conversational interfaces on websites.
  • This article provides more details on specifically how to tie conversational interfaces in to the UX.
  • Specifically, by Identifying Visitor Profiles to segment traffic through tacit (revealed) or explicit (expressed) means.
  • This approach to personalization requires the development of resources to generate user-specific content, such as imagery and user scenarios.
  • And using the combination of those things to offer a personalized experience.

Below : A Personalised Image Of A Bloke Like Me in A BMW X5.

Since the goal of ‘Evaluate’ is to secure a sale by providing users everything they need to exceed their emotional and logical purchasing thresholds, in my view, all brochures should be personalized.

Personalizing Evaluate Materials is already much, much, easier using Generative AI. This article shows how to segment traffic based on conversational inputs and use it to personalize the ‘Evaluate’ element of the website experience.

These Are The Options For Users (including agents) Interacting With Websites

Conversational interfaces are going to be important and will likely become common, soon. However, a conversation isn’t always the best way to get something done. The user’s preferences and context both have an influence on the interaction methodology that the user prefers. Very soon, most websites will offer the combination of the following options:

  • Speech: 
    Verbally spoken interaction for hands-free scenarios or simply when the user prefers it.
  • Typed Chat: 
    Conversationally written text interaction for text-based communication.
  • WIMP: 
    Traditional WIMP (Window, Icon, Menu, Pointer) interfaces for detailed tasks.

There’s not much between a typical business and putting those interfaces in place. Adding the conversational component is as simple as adding one of those ‘chat’ buttons at the bottom of the page which is tied in to the rest of the site. 

I recently implemented a chatbot on my Business Website AI T0ken. Adding the conversational element took all of half an hour, including writing the answers do common visitor FAQs. Having done similar projects elsewhere, building your own chatbot and putting it in place could probably be done for $2k.

Adding conversational interfaces to your website is now very easy.

Adding a conversational chatbot to your site is now very easy. This one took half an hour.

Visit AIT0ken.

My view is that, websites should always offer users the option to choose their preferred interaction method (WIMP, Conversational, Agent). It not only respects user autonomy but also accommodates diverse needs and preferences.

If, instead of the user themselves, their AI Agent is acting on the user’s behalf, they may well already know what sort of a shopper the user is. This could be some of the information the user is happy to share along with simple details like their name and color preferences.

There’s a lot going on there so I have made a diagram.

And of course, these are not the only potential interaction methods available live on the site. It’s possible the product involved is high value. Maybe the user has come to the site to book an appointment with a real human, to discuss their needs. There should be a phone number and a ‘Contact Us’ form too.

So, very soon, there will be more ways of interacting with websites. One of the benefits of that is that businesses now have more information about their users available to them. LLMs and other AI APIs have also made working with AI services much easier. So, how can those things be tied together to personalize the ‘evaluate’ section of a website?

Shopper Archetypes & How They Can Be Used To Personalise The User’s Experience

Research conducted by Leo Burnett, an Australian Advertising agency, in the early 2010s, classified online shoppers as one of 6 different types of Digital Shopper, as part of their study “Peopleshop.” 

Each Digital Shopper archetype has unique preferences, (see table below) to which websites can cater by analyzing user data like browsing patterns, purchase history, and data from their profile taken from previous interactions. I asked ChatGPT to help me determine tacit ways that user behavior on the site could be used to determine which the closest shopper archetype a visitor might match – the answers are shown in the final column.

Below : Leo Burnett’s Shopper Archetypes, What They Want From Your Site & How That Information Could Be Gathered Tacitly

How To Tell, Tacitly, What Type Of Shopper You’re Dealing With

  • Browsing Patterns: 
    Track which pages or products they spend the most time on.
  • Purchase History: 
    Review past purchases to identify trends in preferences or spending habits.
  • Search Queries: 
    Analyze the types of products or information they search for.
  • Interaction with Deals: 
    Observe if they respond more to discount offers, detailed product information, or new trends.
  • Feedback and Reviews: 
    Consider their comments or ratings on products.
  • Social Media Interaction: 
    Review their social media activity related to shopping or products.
  • Text: 
    If they have engaged conversationally, it’s not hard to analyze the content they’ve provided (text, language and phraseology) to estimate what segment they fall in to.

Of course, the user may also explicitly  specify themselves their own criteria for a decent response. 

The idea of categorizing shoppers into different archetypes isn’t entirely new; businesses have long used customer segmentation and profiling to tailor their marketing strategies. In reality, you have only ever been a member of a segment, with a propensity to buy each product they sell, to any company you’ve dealt with. 

What new AI models deliver, however, is a way to explire how AI and data analytics are used to dynamically identify and cater to these shopper types in real-time on digital platforms.

How Can The Information Brands Have And The Segments We Know, be used to Personalize The Experience?

The cornerstone of providing a tailored online shopping experience is the ability to understand these archetypes either by securing the information explicitly or tacitly. By analyzing browsing patterns, purchase histories, and interactions with various website elements, websites can craft a personalized experience that resonates with each shopper type. This approach is not just about making sales but creating a human-centric shopping environment that fosters brand loyalty and repeat business.

To make that a bit clearer, it is entirely possible, to get a Google Analytics report set up which tracks these variables and simply send that report to ChatGPT and use it as the personalization engine which decides what creatives get produced or shown to the user. I’ve actually done something very similar to this on my site hotcrazydating.com.

A Solid Example Of Personalising An Evaluate Experience

I’m a Habitual Sprinter. I just bought a second hand BMW X5. Imagine I had a chat with the chatbot on the dealer’s website and I said this :  Hi, I’m a 50-year-old Australian man, looking to buy a BMW. I know exactly what I want: a BMW x5, in black. I’ve been researching for a while now and I’m ready to make a quick decision. I’m not interested in browsing through endless options or customizing too much. Can you help me with that? I just want the standard features, thanks. I want this to be simple and fast. I’ve read the reviews, compared models, and I know the x5 is the car for me. As long as it drives well and gets me from A to B efficiently, I’m happy. I don’t need any fancy tech or gadgets. Just a reliable, high-performance vehicle. How soon can I get it delivered?

(By the way, that is exactly the sort of thing I would say.)

Now, we feed that in to ChatGPT with a prompt. I’ve just used the live version of ChatGPT 4.0 here but of course, one could achieve the same thing using an API call.

The text is provided to ChatGPT – and a segment is identified. I am a Habitual Sprinter. If you type that text in, as above, you’ll get the same response. If you give it 300 words of your own text, it’ll tell you what sort of a shopper you are.

Now, if the user isn’t ready to buy, the chatbot can offer the user (in this case me) a brochure to assist in the evaluation phase of the purchase. Here’s the rest of the conversation with ChatGPT. And of course, if it had the assets, it could create the brochure. The last point on this list is my favorite.

Me To Chat GPT: Knowing that the user is a Habitual Sprinter, if you were designing a brochure to help them evaluate the purchase, what would you include?

ChatGPT

  • Designing a brochure for a “Habitual Sprinter” shopper, such as the one described, requires a focus on efficiency, straightforward information, and a no-nonsense approach. Here’s what to include:
  • Clear Product Specifications:
    Include detailed but concise information about the BMW X5, focusing on key specifications like engine performance, fuel efficiency, and safety features. Habitual Sprinters appreciate clear, factual data that helps them make quick decisions.
  • Standard Features Highlight:
    Emphasize the standard features of the BMW X5, since the shopper is not interested in customizations or optional extras. Highlight aspects like the standard interior design, infotainment system, and any included driver-assistance technologies.
  • Comparative Analysis: 
    Provide a brief comparison with similar models, focusing on how the X5 stands out. This caters to the shopper’s research-oriented approach, affirming their decision without overwhelming them with choices.
  • Pricing and Financing Options: 
    Clearly outline the pricing for the standard BMW X5 model, along with any financing options or warranties. Habitual Sprinters appreciate having this information upfront to expedite the decision-making process.
  • Testimonials and Reviews: 
    Include a section with customer testimonials and expert reviews that reinforce the reliability and performance of the X5. This helps validate the shopper’s research and decision.
  • Delivery Time Information: 
    Since the shopper is interested in quick delivery, provide clear information on delivery timelines and processes.

By focusing on these elements, the brochure would cater to the efficiency and decisiveness characteristic of a Habitual Sprinter, facilitating a quick and informed purchase decision.

<That’s the end of the text ChatGPT produced in response to the query.>

Just for fun, I gave DALL – E a picture of me and asked it to create a personalized image of me driving a BMW x5.

Personalised brochures will be more prevalent, soon.

Above left : An image of me. Above right : What DALL – E produced when I asked it to show a personalized image of me driving an X5 for inclusion in the brochure. Not bad for a first go.

Conclusion : The Future of Online Retail: A Collaborative AI Approach

The future of website user engagement lies in understanding and adapting to the user’s context, offering personalized experiences, and investing in the necessary components (creative assets like image, text, information tables and even videos) to create such tailored content. By doing so, websites can not only attract users but also engage them in a meaningful and memorable way.

The Digital Shopper Archetype is only one way to segment users. There are a variety of other models – I’ve been using OCEAN to personalize outputs on another AI website I am working on. In time, it will be worth working through many models to see which one provides the best conversion rate.

‘Evaluate’ is an important part of a website. It’s the sizzle that sells the sausage. Some estimates show that personalization can raise sales 10-20% for brands like BMW. If I was selling cars, I’d want to experiment with that, especially given how easy it now is to do. In the space of the couple of hours it took me to write this article, ChatGPT identified the segment I was in, determined the content to include in a brochure designed for me and produced a bespoke image of ‘someone like me’ driving a car, for which I was in the market.