Frequently Asked Questions
Loading...
Have you ever gone to a web site and been utterly confused as to what you're supposed to do? Have you ever been so frustrated with using a web site that you just gave up and left?
Everyone has had negative experiences with hard to understand web navigation, overly complex forms, and cryptic error messages. From a product development perspective, this kind of bad UX design can:
In contrast, past research has found that good UX design can lead to reduced development and maintenance costs, fewer user errors, higher productivity, lower support costs, and increased web traffic and sales revenue.
User testing is when you have people try out a product, like a website, to see if it's easy and enjoyable to use. It helps product teams catch problems early on and make sure their designs actually work well for the people who will use them.
User testing is also a powerful tool for competitive analysis. By observing users interacting with competitors' products or websites, businesses can identify strengths and weaknesses of their own products compared to their rivals. Competitive user testing can inform strategic design decisions and help a product gain a significant advantage in the market.
User testing is complementary to quality assurance (QA) testing. QA testing primarily focuses on identifying technical bugs, functionality errors, and compliance with specifications. In contrast, user testing evaluates the product from the perspective of actual users, uncovering usability issues, pain points, and areas where the design may not meet user needs or expectations. While QA ensures the product works correctly, user testing ensures it works well for the people who will use it.
The main challenges with user testing are that it can be slow, expensive, and does not scale well. Running a standard in-person user study can cost over $10k and take several weeks to complete. Remote user testing is a less expensive alternative, with a study with just 5 participants costing over $1000 and taking a few days to complete. In both cases, the user study is limited in the range of tasks and web pages that can be feasibly tested. Furthermore, the financial cost and amount of time it takes to run these studies makes it hard for teams to iterate quickly and be agile.
Another challenge with user testing is that there are many product teams that have no experience with user testing, making it hard for them to get useful feedback.
Our software for AI user testing simulates people using your web site, making it faster and cheaper to find things that your customers will find confusing and hard to use. You can also run our AI user tests with different personas and then summarize the results, making it fast and easy to find common usability issues across your web site. Lastly, our AI user tests offer predicted metrics such as the SUS (the System Usability Scale) and a Loyalty score (how likely customers are to recommend the web site to others), helping product teams track their improvement over time.
Here is an example of our AI user testing:
AI user testing can help product teams that have little or no UX support. For example, there are many internal intranet sites that have many usability problems but little or no user testing done on them.
AI user testing can also help product managers doing competitive analysis or wanting to improve the quality and usability of their web sites. Our AI user testing can make it easier to flag problems that need to be fixed, prioritize those problems, and track estimated usability and loyalty metrics over time.
Lastly, AI user testing can help Quality Assurance teams in making sure that web sites are up to desired standards and work correctly for customers.
Our AI user testing is not meant to be a complete replacement for user testing. Where possible, product teams should still do some user testing with actual users.
Our value proposition here is that there are many basic usability problems on most web sites, and our AI user testing can find many of these in a fast and cheap manner.
Here are actual examples of usability problems our AI user testing has found:
Please contact us at sales@fuguux.com.
Currently, our software only works on web sites.
Yes, we have several default personas that you can choose from that vary in age, gender, and general background. These personas are currently fictional.
No, and while there is some promising research in this space, we think that this kind of prediction is beyond the current state of the art in AI.
Instead, our AI user testing is good at finding common usability problems in a manner that is fast, cheap, and can scale up for lots of tasks and lots of web pages.
Not out of the box, but we can support tasks that require logins if you contact us at support@fuguux.com.
At a very high level, you give us a web site and a task. We then create an AI agent, which goes to the specified web site, analyzes the structure and content on the page, determines an action, generates an explanation, and then takes the action. These steps are repeated until the agent believes it has completed the task or runs out of steps. Sometimes, however, the user experience is extremely poor and the agent will simply give up earlier. We then generate a video of the agent doing a run and add subtitles to make it easier to follow what the agent is doing.
If you have multiple agents do the same task on the same web site, we synthesize their results to summarize all of the strengths and weaknesses found.
Besides an email address (for your user account with us) and a method of payment, we do not require any company data for you to run an AI user test. Note that the exception here is if you want to do AI user testing on an internal site of yours, in which case we will need some kind of internal access, such as credentials to a test account or VPN access.
All data we store from AI user tests, such as videos and corresponding summaries, are stored on a secure server with access controls in place.
No, we do not use any customer data to train any AI models. Besides your account information, the only way we use your data is to run AI user tests.
We currently use OpenAI to help analyze web sites, and we are using a version of OpenAI's services that does not use data to train future models.
Our team has extensive experience in studying issues of AI ethics, having published scientific articles as well as teaching classes about Responsible AI. So these are issues we have thought a lot about in our team meetings.
First, our bots do not lead to heavy traffic on a site. There are some AI bots that are rapidly crawling web sites and flooding those sites with requests. Our AI user testing operates at about the speed of a single human user, and so it does not result in an inordinate number of page requests. This also means that our AI bots can't be used for denial of service attacks either.
Second, we have put some constraints on our AI user testing to limit potential negative side effects for web site owners. For example, our AI bots currently don't submit forms or send emails, since that might lead to wasted time by actual humans responding to those requests.
Third, we don't use customer data for any purposes beyond AI user testing. We are also not tracking people or collecting data about what things people around the world are clicking on, searching for, or sharing. Our shared goal at fuguUX is to offer tools that can help product teams dramatically improve the usefulness, usability, and desirability of their web sites.
We have a customer support Slack where you can ask questions.
A fugu, also known as a pufferfish, is a cool and quirky fish found in the ocean. When they get scared or threatened, they can puff up into a spiky ball, making them look much bigger and scarier to predators. Be sure not to eat one, though, unless it's from a highly trained and licensed chef, since some fugu have a deadly poison. We think that they're much safer (and cuter!) as mascots than on your dinner plate.
One of the metrics our AI user tests report on is an estimate of the System Usability Scale. SUS ranges from 0 to 100, with 0 being low usability and 100 being high. This scale was developed in the mid 1980s, has been widely used in usability testing, and gives a quantitative way of tracking improvements over time.
So we think that SUS is not sus.
The two founders of fuguUX are Dr. Jason Hong and Dr. Sauvik Das, both professors at the renowned Human-Computer Interaction Institute (HCII) at Carnegie Mellon University (CMU). For those unfamiliar, CMU is widely regarded as one of the top universities in the world for computer science, and the HCII one of the top departments for UX design and the broader field of Human-Computer Interaction.
Jason Hong has over 20 years of experience in teaching and conducting research on UX design, mobile computing, usable privacy and security, and Responsible AI. Hong is a member of CHI Academy ("an honorary group of individuals who have made substantial contributions to the field of human-computer interaction") and an ACM Fellow ("top 1% of ACM members for their outstanding accomplishments in computing and information technology"). He is co-author of The Design of Sites, a popular book on web site design patterns. He helped co-found CMU's Master's of Product Management, a joint program between computer science and the Tepper School of Business. He previously co-founded Wombat Security Technologies, which offered cybersecurity training as a service and was acquired in 2018. Jason's research has been featured in the New York Times, MIT Tech Review, CBS, CNN, Slate, and more. Jason has also given talks at Google, Meta, Samsung, the Congressional Internet Caucus, the World Economic Forum, RSA Conference, and more.
Sauvik Das is an assistant professor of computer science at Carnegie Mellon University. He received his PhD in Human-Computer Interaction from Carnegie Mellon University. His work lies at the intersection of HCI, AI and cybersecurity, and his papers have won several best paper and honorable mention awards at top scientific venues. Das has experience consulting for or working at Twitter, Facebook, Google, and Microsoft Research.
We're located in Pittsburgh PA.