In Part 1 of this guide, we covered the first four phases of audience research: the client discovery interview, mining existing customer data, analyzing the competitive landscape, and defining audience demographics. If you haven’t read it, start there.
This part picks up where the demographics leave off. Knowing who someone is tells you what design constraints to work within. Knowing how they behave, what frustrates them, and what they’re trying to accomplish tells you what the design actually needs to do.
Phase 5: Study How the Audience Behaves
Demographic data tells you who someone is. Behavior data tells you what they actually do on the site, which is often very different. Nielsen Norman Group distinguishes between attitudinal research (what people say) and behavioral research (what people do). Behavioral data produces more reliable design signals.
Set up these tools before or immediately after a site launches.
- Google Analytics 4: Track engagement rate (sessions lasting 10 or more seconds, involving 2 or more page views, or triggering a conversion event), pages per session, top exit pages, and device breakdowns. A large gap between mobile and desktop engagement rates signals a mobile UX problem. Per Google’s own research, the probability of a user leaving increases 32% when page load time goes from 1 second to 3 seconds.
- Microsoft Clarity (free, no traffic limits): Install for heatmaps and session recordings. Click heatmaps reveal which elements users expect to be interactive but aren’t. Scroll heatmaps show whether CTAs are being seen before users leave the page. Session recordings expose rage clicks, form abandonment, and navigation dead-ends.
- On-page surveys: Use Hotjar or Typeform to ask one question at exit intent. “What almost stopped you from contacting us today?” generates qualitative insight that no analytics platform can match. Even 20 to 30 responses can change how a page is structured.
Behavioral research makes the design audit after launch as important as the research done before it.
Phase 6: Identify Pain Points
Pain points are the gaps between what an audience expects and what they’re actually getting. Designing against real pain points means the site addresses actual objections rather than imagined ones.
Nielsen Norman Group categorizes pain points at three levels: interaction problems (a confusing form field), journey problems (a prospect can’t find enough information to feel confident reaching out), and relationship problems (the site looks untrustworthy before a visitor reads a single word). All three levels show up in the design.
Practical methods for surfacing them:
- Mine online reviews. Read the most recent 50 reviews of the client’s business and their top 3 competitors across Google, Yelp, and any industry-specific platforms. Tag each review by category (communication, pricing, quality, speed, trust) and count frequency. What appears most often is most important. UX researchers consistently identify review mining as one of the most underused free sources of authentic audience data.
- Analyze support history. Review the last 50 email or chat support exchanges. Questions asked repeatedly before a sale point to content missing from the website. Complaints after a sale point to expectations the site set incorrectly.
- Run 5-minute customer interviews. Ask three questions: “What were you looking for when you found us?” “What almost stopped you from choosing us?” and “Is there anything you wish you’d known before contacting us?” The answers rewrite homepage copy, FAQ sections, and service page structure more effectively than any design trend.
Pain points addressed directly through content and design are what separate a site that builds trust before the first conversation from one that raises questions it never answers.
Phase 7: Build a User Persona
A user persona turns all of this research into a single reference point for every design decision the project involves. It’s not a fictional character someone invents; it’s a synthesis of real patterns from real research.
According to HubSpot, companies with documented personas are twice as likely to exceed revenue goals, and persona-driven websites are 2 to 5 times more effective for target users. In a Thomson Reuters case study, implementing personas drove a 175% increase in marketing revenue and reduced average conversion time by 72%.
A well-built persona contains a fictional name and representative image; age, location, job title, and income range; device and platform preferences; core goals and what success looks like from their perspective; the pain points they bring to the buying decision; and a short scenario describing how they’d encounter and navigate the site. Research from Delve.ai shows that 3 to 4 personas typically account for over 90% of a company’s revenue, so depth matters more than coverage.
Build the persona after completing phases 1 through 6. Validate it against GA4 data. If the persona describes a 38-year-old mobile-first user, the demographics report should confirm it. Update it after launch as real behavior data accumulates. If you’re working with a design team on a new site or a rebuild, this persona document is what the entire project should be built against.
What the Research Does to Design Decisions
Once the audience is defined, decisions that felt arbitrary become defensible.
- Content hierarchy follows the questions the audience actually asks, in the order they ask them, not the internal logic of the business.
- Navigation labels use the audience’s vocabulary, not industry terminology or internal naming conventions.
- CTA copy reflects the user’s goal (“Get My Free Quote”) rather than the business’s action (“Submit Form”). Placement is validated by heatmap data, not convention.
- Visual design reflects the credibility standards the audience holds for that industry. Research from Kinesis and CXL shows 75% of users judge credibility by visual design, and those judgments form in about 0.05 seconds.
- Social proof matches the audience’s peer group: local names and faces for local service businesses, industry credentials for professional services, case studies that mirror the reader’s situation.
- Page speed is a design requirement. Baymard Institute and Google research consistently show that 40% of users abandon sites that take more than 3 seconds to load. A site that loads in 1 second converts at nearly three times the rate of one that loads in 5.
A website built on this kind of research gives you defensible answers when a client asks why a decision was made. It also performs better from launch, because the decisions were made for the right people in the first place.
Final Thoughts: Research Is What Makes Design Defensible
Every design decision either has a reason or it has a guess. Research converts guesses into reasons, and reasons are what make a website perform instead of just existing.
Not every project will complete all seven phases. But even three of these steps (a structured discovery interview, a review mining session, and a GA4 demographics check) will produce a better result than a site built entirely from assumptions about who the customers are.
The websites that consistently bring in business are the ones where someone did this work before the first page was laid out.