How AI is Augmenting, Not Replacing, UX Researchers
Despite AI being the buzzword in UX right now, there are still lots of concerns about how it’s going to impact research roles. One of the biggest concerns we hear is: is AI just going to replace UX researchers altogether?
The answer, in our opinion, is no. The longer, more interesting answer is that AI is fundamentally transforming what it means to be a UX researcher, and in ways that make the role more strategic, more impactful, and more interesting than ever before.
What AI Actually Does for Research
A 2024 survey by the UX Research Collective found that 68% of UX researchers are concerned about AI's impact on their roles. The fear makes sense, we've all seen how automation has transformed other industries. But what's actually happening is that rather than AI replacing researchers, it's eliminating the parts of research that researchers hate most.
According to Gartner's 2024 Market Guide for User Research, AI tools can reduce analysis time by 60-70%, but not by replacing human insight. Instead, they handle:
Pattern Recognition at Scale: AI can process hundreds of user interviews and identify recurring themes in hours. For a human researcher that same work would take weeks. But those patterns will need human validation because AI doesn't understand why those patterns matter. That's where researchers will continue to add value, and we would argue, become more important than ever.
Synthesis Acceleration: According to research by the Nielsen Norman Group, AI can generate first-draft insight summaries 10x faster than humans. But these summaries still need researcher oversight to ensure context, accuracy, and actionable insights aren't lost.
Multi-language Analysis: AI can analyze feedback in 50+ languages simultaneously, democratizing global research. But cultural context and nuanced interpretation still require human understanding.
Always-On Insights: Traditional research is limited by human availability. Tools like AI interviewers can run 24/7 while your team sleeps, allowing you to get continuous, high-quality user insights.
AI is Elevating the Role of Researchers
We think that what AI is actually doing is making UX researchers more important, not less. By automating the less sophisticated aspects of research, AI is pushing researchers toward the strategic work that only humans can do.
From Operators to Strategists: McKinsey's 2024 research shows that teams using AI research tools spend 45% more time on strategic planning and only 20% on execution, compared to 30% strategy and 60% execution for traditional teams.
From Reporters to Storytellers: With AI handling data processing, researchers can focus on crafting compelling narratives.
From Analysts to Advisors: When freed from manual analysis, researchers become embedded strategic partners.
Human + AI Collaboration
The most effective research teams aren't choosing between human or AI, they're creating collaborative workflows that incorporate AI to augment researchers roles, not replace them:
AI-Powered Data Collection: Automated transcription, sentiment analysis, and preliminary coding happen in real-time during user sessions.
Human-Led Interpretation: Researchers review AI-generated insights, add context, challenge assumptions, and identify what AI might have missed.
Collaborative Synthesis: AI suggests patterns and themes; researchers validate, refine, and connect to business context.
Human Storytelling: Researchers craft narratives, implications, and recommendations that AI cannot generate.
Is it likely that with AI more and more research tasks will become automated? Absolutely. Basic transcription, preliminary coding, and simple pattern recognition are already AI’s bread and butter. But research has never been about these tasks, it's been about understanding users and driving better decisions and that should always be left to humans.
The researchers thriving in 2025 and beyond aren't fighting AI, they're embracing it. They're using AI to handle the tedious 40% of their job so they can focus on the strategic 60% that creates real business value. You have a choice. You can choose to adopt AI as a tool to elevate your role, or you can view it as a threat and get left behind. Our customers tell us that the researchers choosing elevation are finding their roles more strategic, more impactful, and more essential to product success than ever before.
AI isn't replacing UX researchers. It's freeing them to do what they've always done best, understand humans and help build better products. And in a world drowning in data but starving for insight, that human expertise has never been more valuable.
Most product teams treat UX research as something that happens to them: a necessary evil that slows things down or a luxury they can't afford. The best product teams flip this narrative completely. Their research doesn't interrupt their roadmap; it powers it.
"We need insights by Friday."
"Proper research takes at least three weeks."
This conversation happens in product teams everywhere, creating an eternal tension between the need for speed and the demands of rigor. But what if this debate is based on a false choice?
Research that Moves at the Speed of Product
Product development has accelerated dramatically. Two-week sprints are standard. Daily deployment is common. Feature flags allow instant iterations. In this environment, a four-week research study feels like asking a Formula 1 race car to wait for a horse-drawn carriage.
The pressure is real. Product teams make dozens of decisions per sprint, about features, designs, priorities, and trade-offs. Waiting weeks for research on each decision simply isn't viable. So teams face an impossible choice: make decisions without insights or slow down dramatically.
As a result, most teams choose speed. They make educated guesses, rely on assumptions, and hope for the best. Then they wonder why features flop and users churn.
The False Dichotomy
The framing of "speed vs. rigor" assumes these are opposing forces. But the best research teams have learned they're not mutually exclusive, they require different approaches for different situations.
We think about research in three buckets, each serving a different strategic purpose:
Discovery: You're exploring a space, building foundational knowledge, understanding thelandscape before you commit to a direction. This is where you uncover the problems worth solving and identify opportunities that weren't obvious from inside your product bubble.
Fine-Tuning: You have a direction but need to nail the specifics. What exactly should this feature do? How should it work? What's the minimum viable version that still delivers value? This research turns broad opportunities into concrete solutions.
Delivery: You're close to shipping and need to iron out the final details: copy, flows, edge cases. This isn't about validating whether you should build it; it's about making sure you build it right.
Every week, our product, design, research and engineering leads review the roadmap together. We look at what's coming and decide which type of research goes where. The principle is simple: If something's already well-shaped, move fast. If it's risky and hard to reverse, invest in deeper research.
How Fast Can Good Research Be?
The answer is: surprisingly fast, when structured correctly!
For our teams, how deep we go isn't about how much time we have: it's about how much it would hurt to get it wrong. This is a strategic choice that most teams get backwards.
Go deep when the stakes are high, foundational decisions that affect your entire product architecture, things that would be expensive to reverse, moments where you need multiple stakeholders aligned around a shared understanding of the problem.
Move fast when you can afford to be wrong, incremental improvements to existing flows, things you can change easily based on user feedback, places where you want to ship-learn-adjust in tight loops.
Think of it as portfolio management for your research investment. Save your "big research bets" for the decisions that could set you back months, not days. Use lightweight validation for everything else.
And while good research can be fast, speed isn't always the answer. There are definitely situations where deep research needs to run and it takes time. Save those moments for high stakes investments like repositioning your entire product, entering new markets, or pivoting your business model. But be cautious of research perfectionism which is a risk with deep research. Perfection is the enemy of progress. Your research team shouldn’t be asking "Is this research perfect?" but instead "Is this insight sufficient for the decision at hand?"
The research goal should always be appropriate confidence, not perfect certainty.
The Real Trade-Off
The choice shouldn’t be speed vs. rigor, it's between:
Research that matters (timely, actionable, sufficient confidence)
Research that doesn't (perfect methodology, late arrival, irrelevant to decisions)
The best research teams have learned to be ruthlessly pragmatic. They match research effort to decision impact. They deliver "good enough" insights quickly for small decisions and comprehensive insights thoughtfully for big ones.
Speed and rigor aren't enemies. They're partners in a portfolio approach where each decision gets the right level of research investment. The teams winning aren't choosing between speed and rigor—they're choosing the appropriate blend for each situation.
Let's talk about something that's changing the game for all of us in digital product design: AI search. It's not just a small update; it's a complete revolution in how people find information online.
Today's AI-powered search tools like Google's Gemini, ChatGPT, and Perplexity AI aren't just retrieving information they're having conversations with users. Instead of giving you ten blue links, they're providing direct answers, synthesizing information from multiple sources, and predicting what you really want to know.
This raises a huge question for those of us creating digital products: How do we design experiences that remain visible and useful when AI is deciding what users see?
AI Search Is Reshaping How Users Find and Interact with Products
Users don't browse anymore: they ask and receive. Instead of clicking through multiple websites, they're getting instant, synthesized answers in one place.
The whole interaction feels more human. People are asking complex questions in natural language, and the AI responses feel like real conversations rather than search results.
Perhaps most importantly, AI is now the gatekeeper. It's deciding what information users see based on what it determines is relevant, trustworthy, and accessible.
This shift has major implications for product teams:
If you're a product manager, you need to rethink how your product appears in AI search results and how to engage users who arrive via AI recommendations.
UX designers—you're now designing for AI-first interactions. When AI directs users to your interfaces, will they know what to do?
Information architects, your job is getting more complex. You need to structure content in ways that AI can easily parse and present effectively.
Content designers, you're writing for two audiences now: humans and AI systems. Your content needs to be AI-readable while still maintaining your brand voice.
And UX researchers—there's a whole new world of user behaviors to investigate as people adapt to AI-driven search.
How Product Teams Can Optimize for AI-Driven Search
So what can you actually do about all this? Let's break it down into practical steps:
Structuring Information for AI Understanding
AI systems need well-organized content to effectively understand and recommend your information. When content lacks proper structure, AI models may misinterpret or completely overlook it.
Key Strategies
Implement clear headings and metadata – AI models give priority to content with logical organization and descriptive labels
Add schema markup – This structured data helps AI systems properly contextualize and categorize your information
Optimize navigation for AI-directed traffic – When AI sends users to specific pages, ensure they can easily explore your broader content ecosystem
LLM.txt Implementation
The LLM.txt standard (llmstxt.org) provides a framework specifically designed to make content discoverable for AI training. This emerging standard helps content creators signal permissions and structure to AI systems, improving how your content is processed during model training.
How you can use Optimal: Conduct Tree Testing to evaluate and refine your site's navigation structure, ensuring AI systems can consistently surface the most relevant information for users.
Optimize for Conversational Search and AI Interactions
Since AI search is becoming more dialogue-based, your content should follow suit.
Write in a conversational, FAQ-style format – AI prefers direct, structured answers to common questions.
Ensure content is scannable – Bullet points, short paragraphs, and clear summaries improve AI’s ability to synthesize information.
Design product interfaces for AI-referred users – Users arriving from AI search may lack context ensure onboarding and help features are intuitive.
How you can use Optimal: Run First Click Testing to see if users can quickly find critical information when landing on AI-surfaced pages.
Establish Credibility and Trust in an AI-Filtered World
AI systems prioritize content they consider authoritative and trustworthy.
Use expert-driven content – AI models favor content from reputable sources with verifiable expertise.
Provide source transparency – Clearly reference original research, customer testimonials, and product documentation.
Test for AI-user trust factors – Ensure AI-generated responses accurately represent your brand’s information.
How you can use Optimal: Conduct Usability Testing to assess how users perceive AI-surfaced information from your product.
The Future of UX Research
As AI search becomes more dominant, UX research will be crucial in understanding these new interactions:
How do users decide whether to trust AI-generated content?
When do they accept AI's answers, and when do they seek alternatives?
How does AI shape their decision-making process?
Final Thoughts: AI Search Is Changing the Game—Are You Ready?
AI-powered search is reshaping how users discover and interact with products. The key takeaway? AI search isn't eliminating the need for great UX, it's actually making it more important than ever.
Product teams that embrace AI-aware design strategies, by structuring content effectively, optimizing for conversational search, and prioritizing transparency, will gain a competitive edge in this new era of discovery.
Want to ensure your product thrives in an AI-driven search landscape? Test and refine your AI-powered UX experiences with Optimal today.
The footer of a website sits at the very bottom of every single web page and contains links to various types of content on your website. It’s an often overlooked component of a website, but it plays several important roles in your information architecture (IA) – it’s not just some extra thing that gets plonked at the bottom of every page.
Getting your website footer right matters!
The footer communicates to your website visitors that they’ve reached the bottom of the page and it’s also a great place to position important content links that don’t belong anywhere else – within reason. A website footer is not a dumping ground for random content links that you couldn’t find a home for, however there are some content types that are conventionally accessed via the footer e.g., privacy policies and copyright information just to name a few.
Lastly, from a usability and navigation perspective, website footers can serve as a bit of a safety net for lost website visitors. Users might be scrolling and scrolling trying to find something and the footer might be what catches them and guides them back to safety before they give up on your website and go elsewhere. Footers are a functional and important part of your overall IA, but also have their own architecture too.
Read on to learn about the types of content links that might be found in a footer, see some real life examples and discuss some approaches that you might take when testing your footer to ensure that your website is supporting your visitors from top to bottom.
What belongs in a website footer
Deciding which content links belong in your footer depends entirely on your website. The type of footer, its intent and content depends on its audience of your customers, potential customers and more — ie your website visitors. Every website is different, but here’s a list of links to content types that might typically be found in a footer.
Legal content that may include: Copyright information, disclaimer, privacy policy, terms or use or terms of service – always seek appropriate advice on legal content and where to place it!
Your site map
Contact details including social media links and live chat or chat bot access
Customer service content that may include: shipping and delivery details, order tracking, returns, size guides, pricing if you’re a service and product recall information.
Website accessibility details and ways to provide feedback
‘About Us’ type content that may include: company history, team or leadership team details, the careers page and more
Key navigational links that also appear in the main navigation menu that is presented to website visitors when they first land on the page (e.g. at the top or the side)
Website footer examples
Let’s take a look at three diverse real life examples of website footers.
IKEA US
IKEA’s US website has an interesting double barrelled footer that is also large and complex – a ‘fat footer’ as it’s often called – and its structure changes as you travel deeper into the IA. The below image taken from the IKEA US home page shows two clear blocks of text separated by a blue horizontal line. Above the line we have the heading of ‘All Departments’ with four columns showing product categories and below the line there are seven clear groups of content links covering a broad range of topics including customer service information, links that appear in the top navigation menu and careers. At the very bottom of the footer there are social media links and the copyright information for the website.
IKEA US home page footer (accessed May 2019)
As expected, IKEA’s overall website IA is quite large, and as a website visitor clicks deeper into the IA, the footer starts to change. On the product category landing pages, the footer is mostly the same but with a new addition of some handy breadcrumbs to aid navigation (see below image).
IKEA US website footer as it appears on the product category landing page for Textiles & Rugs (accessed May 2019).
When a website visitor travels all the way down to the individual product page level, the footer changes again. In the below image found on the product page for a bath mat, while the blue line and everything below it is still there, the ‘All Departments’ section of the footer has been removed and replaced with non-clickable text on the left hand side that reads as ‘More Bath mats’ and a link on the right hand side that says ‘Go to Bath mats’. Clicking on that link takes the website visitor back to the page above.
IKEA US website footer as it appears on the product page for a bath mat (accessed May 2019).
Overall, evolving the footer content as the website visitor progresses deeper into the IA is an interesting approach - as the main page content becomes more focussed as does the footer while still maintaining multiple supportive safety net features.
M.A.C Cosmetics US
The footer for the US website of this well known cosmetics brand has a four part footer. At first it appears to just have three parts as shown in the image below: a wide section with seven content link categories covering a broad range of content types as the main part with a narrow black strip on either end of it making up the second and third parts. The strip above has loyalty program and live chat links and the strip below contains mostly links to legal content.
MAC Cosmetics US website footer with three parts as it appears on the home page upon first glance (accessed May 2019).
When a website visitor hovers over the ‘Join our loyalty program’ call to action (CTA) in that top narrow strip, the hidden fourth part of the footer which is slightly translucent pulls up like a drawer and sits directly above the strip capping off the top of the main section (as shown in the below image). This section contains more information about the loyalty program and contains further CTAs to join or sign in. It disappears when the cursor is moved away from the hover CTA or it can be collapsed manually via the arrow in the top right hand corner of this fourth part. It’s an interesting and unexpected interaction to have with a footer, but it adds to the overall consistent and cohesive experience of this website because it feels like the footer is an active participant in that experience.
MAC Cosmetics US website footer as it appears on the home page with all four parts visible (accessed May 2019).
Domino’s Pizza US
Domino’s Pizza’s US website has a reasonably flat footer in terms of architecture but it occupies as much space as a more complex or deeper footer. As shown in the image below, its content links are presented horizontally over three rows on the left hand side of the footer and these links are visually separated by forward slashes. It also displays social media links and some advertising content on the right hand side. The most interesting feature of this footer is the large paragraph of text titled ‘Legal Stuff’ below the links. Delightfully it uses direct, clear and plain language and even includes a note about delivery charges not including tips and to ‘Please reward your driver for awesomeness’.
Domino’s Pizza US website footer as it appears on the home page (accessed May 2019).
How to test a website footer
Like every other part of your website, the only way you’re going to know if your footer is supporting your website visitors is if you test it with them. When testing a website’s IA overall, the footer is often excluded. This might be because we want to focus on other areas first or maybe it’s because testing everything at once has the potential to be overwhelming for our research participants.
Testing a footer is fairly easy thing to do and there’s no right or wrong approach – it really does depend on where you are up to in your project, the resources you have available to you and the size and complexity of the footer itself!
If you’re designing a footer for a new website there’s a few ways you might approach ensuring your footer is best supporting your website visitors. If you’re planning to include a large and complex footer, it’s a good idea to start by running an open card sort just on those footer links. An open card sort will help you understand how your website visitors expect those content links in your footer to be grouped and what they think those groups should be called.
If you’re redesigning an existing website, you might first run a tree test on the existing footer to benchmark test it and to pinpoint the exact issues. You might tree test just the footer in the study or you might test the whole website including the footer. Optimal's tree testing is really flexible and you can tree test just a small section of an IA or you can do the whole thing in one go to find out where people are getting lost in the structure. Your approach will depend on your project and what you already know so far. If you suspect there may be issues with the website’s footer, for example, if no one is visiting it and/or you’ve been receiving customer service requests from visitors to help them find content that only lives in the footer, it would be a good idea to consider isolating it for testing. This will help you avoid any competition between the footer and the rest of your IA as well as any potential confusion that may arise from duplicated tree branches (i.e. when your footer contains duplicate labels).
If you’re short on time and there aren’t any known issues with the footer prior to a redesign, you might tree test the entire IA in your benchmark study, iterate your design and then along with everything else, include testing activities for your footer in your moderated usability testing plan. You might include a usability testing scenario or question that requires your participants to complete a task that involves finding content that can only be found in the footer (e.g., shipping information if it’s an ecommerce website). Also keep a close eye on how your participants are moving around the page in general and see if/when the footer comes into play – is it helping people when they’re lost and scrolling? Or is it going unnoticed? If so, why and so on. Talk to your research participants like you would about any other aspect of your website to find out what’s going on there. When resources are tight, use your best judgement and choose the research approach that’s best for your situation, we’ve all had moments where we’ve had to be pragmatic and do our best with what we have.
When you’re at a stage in your design process where you have a visual design or concept for your footer, you could also run a first-click test. First-click tests are quick and easy and will help you determine how your website visitors are faring once they reach your footer and if they can identify the correct content link to complete their task. Studies can be run remotely or in person and just like the rest of the tools in Optimal's user research platform, are super quick to run and great for reaching website visitors all over the world simply by sharing a link to the study.
