uxGPT: Create fake data for wireframes in minutes
Patrick Neeman
UX Leader at Workday. Author of uxGPT: Mastering AI Assistants for UX Designers and Product Managers. Working on great things with Gen AI since November 2022. Ex-Microsoft. Opinions are mine.
Fake data?—?the bane of existence for user experience designers. Creating this data is time-consuming and requires much imagination and testing of different content.
I have spent countless hours repeatedly refining this content to ensure the wireframes represent what the application would display. Besides using Bob Ross Ipsum or the NSFW Samuel L. Jackson Ipsum, I always prefer to simulate the real experience as closely as possible.
Representative data is ridiculously important for several reasons.
It serves as a placeholder that mimics real-world content, helping visualize how the interface will function and behave with actual data. As it simulates how users interact with the product, we assess usability and functionality more realistically.
Fake data identifies potential design flaws early on. By integrating representative content, we can better understand how different data types and volumes affect layout. This enables us to test edge cases and ensure the interface remains intuitive and efficient across various scenarios.
Presenting wireframes with fake data facilitates more meaningful discussions and decisions during design reviews. It shifts the focus from hypothetical situations to concrete examples, prompting productive dialogues about content priorities, information hierarchy, and interaction patterns.
You can’t test your design without having the right representative data in?place.
Now, Chat GPT accelerates this. It saves hours of time and effort and creates more realistic content to test your concepts.
You can start with a prompt and refine it quickly. Chat GPT will create realistic-looking content and ensure the data looks correct, down to addresses and phone numbers that match the right metropolitan areas.
I’m going to show you how to do it.
One extra credit item -- you can export the data to Google Sheets directly from Chat GPT 4o.
Start With A?Prompt
I would start with the following sentence: “Generate a table of realistic fake data with 25 rows for a (topic) application.” It’s a great way to start because Chat GPT will give you a table of information that it believes is appropriate for the topic you select and suggest fields.
The fields don’t have to be right; they give you a baseline. I always state “realistic fake data” as part of the prompt because it generates more representative data.
Example Topics
Prompt
Generate a table of realistic fake data with 25 rows for a customer relationship management application.
Refine Columns
Once you get the result you want from the initial table, you can review it with your stakeholders if you have the right columns. If you don’t, adding them to the prompt is easy. I’ll list the columns and add them below.
Prompt
Create a table of realistic fake data with 25 rows for a customer relationship management application with the following columns: Customer ID, Name, Email Address, Phone, Company, Address, Industry, Revenue, First Contacted, Last Contacted, and Next Follow-Up.
Refine Format
Let’s refine the format.
领英推荐
Fake data in the proper format is crucial because it accurately reflects user behaviors and needs, aligning closely with real-world scenarios. It also gives the engineers the proper context for formatting the data and solves many of the what-it-looks-like questions.
It handles localization very well, with a few examples listed below.
Format Questions
Formats
Prompt
Create a table of realistic fake data with 25 rows for a customer relationship management application with the following columns: Customer ID that’s not sequential, Name, Email Address, Phone in US format or EU format, Company, Industry, Revenue in 7-digit US or German or Russian format with proper localization, First Contacted in YYYY-MM-DD format, Last Contacted in YYYY-MM-DD format, Next Follow-Up in YYYY-MM-DD format.
Export to other formats
You’ll be able to copy and paste what you generated, but there is one other nice feature of Chat GPT – you can export to other formats. Use the prompt below to do it. It is that easy.
Prompt
Export this data to Excel, csv, xml and json.
uxGPT Fake?Data Custom GPT
Don’t want to do the work yourself? Not a problem. I’ve done a lot of the legwork for you.
Try this custom GPT at uxGPT Fake Data to create fake data in seconds.
Articles in the uxGPT series
About Patrick Neeman
Patrick Neeman works at Evisort as the Head of User Experience and Research. He has lead user experience teams at many data companies, like Knowable, Icertis, Apptio, and Jobvite and nPario. You can find his blog Usability Counts here, and he also writes on Medium.
He has taught at General Assembly as a user experience instructor.
All his opinions are his alone.
Other Links
I help startups skyrocket conversions & slash churn in 30 Days—guaranteed. Product Design | Lean Design Systems | Design Ops
8 个月I really like this idea, and have some thoughts to add! When I had a stint as a software engineer, and started working with data mocks so that we could start building front end applications without waiting for backend APIs to be built first, one idea that came to my mind as a designer was: It would help us if we had a term like “data fidelity.” If low fidelity data is lorem ipsum, then I expect an LLM like ChatGPT could generate some good mid-fidelity data. High(er) fidelity data often depicts specific details of a user journey. For example, if one piece in the mock is a result from one or two other pieces of data elsewhere in the workflow, we need to show that eventually, right? As we give demos, it’s helpful to clarify to stakeholders what level fidelity the data is, to manage expectations, as well as convey how far along the feature is in the thought process.