Improving Your Web Shop's UX with ChatGPT 4.0: A Practical Guide
The virtual UX consultant who just joined your team. Image generated by AI with Midjourney.

Improving Your Web Shop's UX with ChatGPT 4.0: A Practical Guide

OpenAI , the company behind ChatGPT, recently released ChatGPT 4.0 - its newest model of the generative AI chat engine which allows it to analyse documents, create pictures with Dall-E and, most importantly for this guide, browse the web with Bing.

Given ChatGPT's extensive knowledge of various best practices in the sales, marketing, and digital product development sector, this is an opportunity for small teams of marketing or product managers to implement UX best practices into their marketing strategies and feature planning processes with a virtual UX consultant.

In this guide, we will learn how to use ChatGPT's newest features to analyse our website and landing pages based on UX best practices with the goal of identifying areas of improvement. We will...

  • understand our target audience's needs and generate customer feedback with a simulated member of our target audience,
  • simulate UX evaluations by a virtual UX consultant, and
  • identify the overlap between the two simulated personas telling us what our customers need and what UX experts recommend.

Important disclaimer: Throughout our journey in this guide, we will not consider any real data on customer behaviour, we will not link our analytics tools, and we will not include information on any special campaigns or company strategies. Although, ChatGPT is helpful for us to identify areas of improvement, it is not yet capable of prioritising in line with our business goals or previously collected data. The AI can only respond to data it was trained on or data we provide. ChatGPT does not replace a truly experienced UX specialist with domain and industry expertise, insights into proprietary analytics data, and a committed company strategy. However, it gives teams and companies without the funds to hire a UX specialist the advantage of guiding technical or business-oriented team members into a good direction towards improving the user experience for their customers.

Getting started

To ensure that you can follow along, you need to sign up for ChatGPT and subscribe to the paid plan: ChatGPT Plus. This subscription enables the newest version 4.0 for your account which includes all the required features for this guide. Previous versions of the tool are not yet able to browse the web and visit links.

In this guide, you will learn how to role play with ChatGPT. We will use this method to learn about customer behaviour and to add a virtual UX consultant to our team. We will provide the AI with sufficient information about our company and our target audience, so that it can simulate the necessary personas and generate recommendations for us. In addition to that, we will learn how to let the tool analyse our website and recommend areas of improvement based on sales, marketing, and UX best practices. In the end, we will identify the overlap between:

  • the customer feedback given by a simulated visitor matching the criteria of our target audience
  • the recommendations of our virtual UX consultant on areas to improve

... to ensure that we can create a plan on how to improve the overall performance of our website and landing pages.

The scenario: eco-friendly engagement rings

Let's assume that our company is based in Germany and we predominantly cater to a german-speaking audience. We manage a web shop specialised in selling engagement rings produced under fair conditions and with ecologically friendly resources. In order to set up the foundation for our analysis, we let ChatGPT describe the target audience by providing it with some context:

What is a suitable target audience for a Germany-based company selling engagement rings produced under fair conditions and with ecologically friendly resources? Include demographics, psychographics, and behavioural patterns in your response.

Based on this prompt, ChatGPT gives us a helpful description of a suitable target audience for our company. This information will be stored in the AI's memory which means that we will be able to reference it in future messages. Keep in mind that providing more information about the company, marketing strategies and the products leads ChatGPT to describe more specific target audiences. The more effort you invest into providing data to the tool, the more specific the generated outcome will be.

Demographics, psychographics and behavioural patterns can give you and the tool a great idea of how a member of the target audience would react to certain elements on a website.

You should iterate through this process until you feel like the tool has a good understanding of what the most suitable target audience for your business is. If you are unaware of your target audience, then this tool can help you with identifying suitable candidates. Be sure to address shopping behaviour, demographics, and psychographics. The more detailed your input, the better ChatGPT's assumption of the following personas.

Understanding customer needs and initial visit perception

Now that ChatGPT is aware of the business and our target audience, it is time for us to start role-playing. In the next step, we let ChatGPT assume the role of a member of our target audience. Based on previous messages, the AI will create a persona for itself and act in its role. We will use this feature to simulate a member of our highly-specific target audience and let them analyse our web shop and landing pages. Our goal is to collect feedback on the attractiveness of the offered product with the expectation for the AI to provide feedback on how likely it is to complete certain actions.

In this example, we tell ChatGPT to assume the role of a persona - an imaginary customer - a 32-year old woman who is planning to get engaged to her female long-term partner - with this prompt:

Assume the identity of a member of the described target audience with the following additional properties: You are a 32-year old woman planning to get engaged to her female long-term partner. In your newly assumed role, visit [URL of the web shop] and give me feedback on how the website could be improved to make you more likely to "buy an engagement ring", "sign up to the newsletter", "click through the catalog of products", "read more about the company", and "recommend the web shop to a friend".

Add your key conversion actions to the end of the prompt and ChatGPT will give recommendations on what to improve to target those specific user actions.

After a thorough analysis of the provided link(s), the tool starts generating feedback based on the provided data on the imaginary customer and considering the previously provided data on our target audience. ChatGPT gives us constructive feedback on a customer's perception of our offerings. Here's what the simulated visitor has to say and notice how the wording underlines how seriously the AI takes its persona:

The AI assumes the role of a customer and projects the assumed values and needs onto the experience it had when visiting the website.

The AI speaks of its values and doubts. It emphasises how a focus on sustainability could deepen the AI's connection with the brand. The AI displays an awareness of the persona being a member of an under-represented minority and underlines how a more inclusive and diverse storytelling approach would make the shopping experience more personal and engaging. In line with this response and based on this analysis, ChatGPT also provides the following recommendations for the requested ideas to improve specific conversions:

ChatGPT recommends incentives which would boost engagement of the simulated customer.

The AI gave us a thorough summary of our website's strengths, areas of improvement, and straight-forward recommendations on how to boost engagement - specific to our desired target audience. This feedback, however, is highly subjective towards the assumed persona. It does not necessarily highlight the most effective areas of improvement, nor does it give us a clear (and proven) strategy on how to improve the user experience to boost conversions.

To ensure that we consider effective best practices, we will simulate a virtual UX consultant in the next step.

Finding an overlap between customer needs and UX best practices

Based on the understanding of our customers' needs, we want to identify areas of improvement for which established best practices can be used to boost engagement. In the next step we will let ChatGPT assume the role of a UX expert to analyse the same links and see which recommendations of the two personas overlap with the following prompt:

Next, assume the role of a UX Consultant hired to improve the user experience of the web shop. Use your understanding of UX best-practices and e-commerce sales methods to recommend areas of improvement for the website [URL of the website]

The AI now switches the persona and analyses the links again. It generates new knowledge based on ChatGPT's extensive learning data on the topic of user experience design. This knowledge is then applied to the analysis of our website and the provided business strategy. The AI provides us with an extensive strategy and detailed steps on what to consider to improve the user experience of the web shop. At this point, the AI's output is very detailed and long. It mentions twelve areas of improvement, each of them listing specific things to change in order to ensure that the website follows established UX best practices. Here's a shortened summary of the feedback for our fair and eco-friendly jewellery shop:

I asked ChatGPT to shorten its response so that it fits into a neat screenshot. The original results was much, much more detailed.

Now that we have identified key areas of improvement from the perspective of a user experience designer, it is time for us to identify an overlap between the simulated customer's needs and the UX consultant's recommendations.

Ideally, this will lead to us improving those UX elements which affect our desired target audience the most. It will ensure that we can follow established best practices while keeping our focus on the customers' needs.

To identify the overlap between both personas, we will let the AI compare the two sets of simulated feedback and identify the top 3 areas of improvement for us to consider next when creating a sales, marketing, or product development strategy for the web shop:

Find an overlap between the visitor's recommendations and the UX expert's recommendations. Identify the top 3 areas of improvement which we should consider next when creating a marketing and product development strategy for the web shop.
A shortened version of the exercise's result. This overlap will help us commit to a marketing and product development strategy.

The resulting response lists those lacking UX elements on our website which will likely have the highest impact towards conversions if improved upon. Based on this recommendation, we should be able to start ideating and testing to see if an improvement in overall performance can be measured.

Improving these elements will positively affect the customers experience when browsing our website - and they are based on established design patterns which are likely to improve performance.

This overlap is a great step towards identifying key features to improve in our web shop.

Conclusion

It is important to say, that these recommendations are based on extremely limited data. The results of this exercise can not (yet) compete with a truly experienced user experience specialist who has access to proprietary customer data, industry knowledge, long-term analytics, company/campaign strategies, and a specific understanding of the product and its users. ChatGPT takes the personas in the role-playing game very seriously and responds with a confidence which could lead us into trusting the recommendations more than we should.

Always remember, a good product manager creates a hypothesis based on an idea and tests it with data. ChatGPT and similar generative AI tools are great at coming up with creative ideas and providing helpful notes from various knowledge spaces. At this technological level, though, you should make sure that you do not blindly follow its recommendations and only use it as a starting point to explore your strategic options.

An android shopping for an engagement ring. Image generated by AI with Midjourney.

I hope you enjoyed this little exercise with ChatGPT 4.0's newest feature. OpenAI has already released the next set of features to paid subscribers. Custom GPTs let users store information and define certain rules for the AI engine to follow. In the context of this guide, this is especially helpful when setting up the tool with proprietary data like user personas, or content strategies.

Let me know what you think in the comments. How did this guide help you see your product websites from a new perspective? Which other roles do you want the AI to assume?

Bonus exercise and more things that ChatGPT can help you with:

There are many other things ChatGPT can already help you with. Here are just a few ideas what you can do with the tool:

  • Prepare a marketing & sales strategy based on your target audience
  • Conduct product discovery based on user behaviour
  • Write convincing copy for your website's landing pages

As a bonus exercise, try simulating a user interview between the simulated target audience member and a UX researcher. Provide both personas with sufficient data and tell the UX researcher to interview the user. If done correctly, ChatGPT will generate an interview between both personas and give you an interesting perspective on the exchange.

Jan Simon

Software Engineering Consultant bei Opencreek | Spezialist für performante, nutzerzentrierte App-Entwicklung

10 个月

This AI article series is gold! Keeping us all in the loop with the latest in AI insights. Can't wait for the next one ??

Sami Hamed

Digitale Innovation & Produktentwicklung durch custom Software L?sungen ?? mit Opencreek Squads

10 个月

Julia Bernhardt have you tried this? It's a neat trick to identify areas of improvement for onsite campaigns and all sorts of landing pages.

回复
Alex Kunkel

Leveraging underprized attention online for funk

10 个月

I‘m really struck by the intelligent use of personas in the process. In performance marketing I often view them as overused but this interview construct makes perfect sense. ??

要查看或添加评论,请登录

Sami Hamed的更多文章

社区洞察

其他会员也浏览了