Just-Enough Research for Solo Designers

The Context Nobody Gives You

The phrase "just enough research" comes from Erika Hall's book of the same name, which treats research as a short cookbook of methods anyone on the team can learn and run. Hall wrote that for a world where most designers had at least one researcher somewhere nearby, which is not the world you and I are in. Mule Books

In 2024, 14% of surveyed organizations had zero dedicated researchers, double the prior year and the highest rate since 2020. A fifth of companies had laid off researchers in the prior year, and 62% of organizations reported no ResearchOps function, up from 50% in 2023. Mid-level designers are suddenly doing research they were never really taught to do, on timelines they do not own, with no one watching for mistakes. user interviews

At the same time, 80% of researchers and designers running research now use AI tools, up more than 20 percentage points year over year. As someone who has helped teams wire AI into research workflows, I can confirm the tooling floor has dropped a lot: solo designers now have AI-moderated interviews, automatic synthesis, free behavioral analytics, and unmoderated testing tools that simply were not practical a few years ago. user interviews CleverX

So the question is not "should we research." It is "how do we do enough, given the team and calendar we actually have."


The Core Problem With Doing Research Alone

Before you pick a method or open a testing tool, solo designers run into one big structural problem. No one is around to push back on your interpretation.

On a two-person team, every hot take about what a user "meant" can get challenged by someone who watched the same session. When you are alone, your starting hypothesis usually survives the whole process, unless the evidence is so obvious that it punches through your story about the design. Most research evidence is not that dramatic. User Intuition

Studies in design and cognitive psychology show that both disconfirming and neutral information get misread as confirming what we already believed, especially when we are tired or personally invested. If you built the prototype and ran the test, you are actually in the worst position to interpret what broke. Not because you are bad at your job, but because you are human. Guerilla Girl ScienceDirect St. Jude

The more time and pride you have sunk into a prototype, the harder it is to accept that the core flow is off. I have watched teams (including my own, in past lives) cling to a flow for weeks longer than the data justified.

So the starting problem for solo work is not "how do I find participants" or "which method is cool." It is "how do I keep myself from believing the first story my brain wants to tell about this data."

The Three Structural Gaps Solo Research Faces

Gap What It Normally Does What Replaces It Solo
Second reviewer Pushes back on interpretation, catches bias Simple analysis protocol, short review session with a teammate, or intentionally adversarial prompts
Research repository Prevents duplicate questions, builds cumulative insight Personal Notion or Airtable log with tags and a 30-minute intake per study
Methodological oversight Catches leading questions, validates screeners A few solid templates, pre-tested scripts, and borrowing from public templates (NNG, Krug, etc.)

The "Just Enough" Principle, Sharpened

Hall's core idea still holds. Just enough research means "enough to make this specific decision with acceptable confidence, given what it costs to be wrong."

The move that saves the most pain is boring but powerful: clarify your goal before you touch a method. Mule Design has a nice way of putting it: know how and when you will use the insight before you plan any research. Write down your research question before you pick activities. A lot of solo research dies quietly because the question was fuzzy, so the findings could support absolutely anything. Mule Design

A simple 2×2 helps you right-size effort. Plot your current question by:

Risk of being wrong
Go research-heavy Generative work: JTBD-style interviews, diary studies, contextual inquiry.
Design with guardrails Evaluative checks: prototype tests, click tests, comprehension checks.
Research-lite Quick tests, analytics reviews, competitor sweeps.
Ship & instrument Ship, instrument, and watch. No formal study needed.
Problem clarity →

Most solo designers I talk to (and work with) default into "high-fidelity prototype and evaluative testing" regardless of the quadrant, because that is the muscle they have. The 2×2 is just there to interrupt autopilot. UX Collective UXR Guild


The Zero-Budget Arsenal

Let us talk about the fun part: doing real research with roughly the budget of two coffees and a bus fare. I have run a lot of these setups with teams that had "no budget" written in bold italics on their roadmap, including some case studies you can find on abseth.com.

Guerrilla Testing

Guerrilla usability testing is walking into public spaces like coffee shops, libraries, coworking spaces, or campus lobbies with a laptop or phone and asking strangers for 5-10 minutes. No scheduling, no incentives, no lab. Maze

Its job is to catch serious usability issues before engineers spend a week baking them into production. It is not meant to be your sole research practice, but it is miles better than shipping based on your own gut. Thinking Design

How to run a coffee shop test without feeling like a street magician:

  1. Pick your spot. Go where people linger. Independent coffee shops and libraries beat train stations nine times out of ten. Please also do not test your very niche B2B compliance dashboard on random students unless they actually look like your users. Bitovi
  2. Approach like a human. Give people a minute to settle, then say something like: "Hey, I am working on a new app and I want to see if it makes sense to regular people, not tech folks. Would you have five minutes to click through it? There are no wrong answers. I am testing the design, not you." That last bit matters more than it should. Koji
  3. Ask for tasks, not opinions. Give people something to do: "Can you try to sign up for a free account?" is good. "What do you think of the design?" is how you collect vague opinions that age poorly in Figma comments.
  4. Use think-aloud. Ask them to say what they are thinking as they go. You will hear all the "Wait, where is X?" moments in real time. This protocol is still the best free research tool out there. Racoonn
  5. Aim for 5-10 short sessions. Ten would be great, five is usually enough to surface the loudest problems for a solo designer on a tight day. Thinking Design

One important caveat. Guerrilla participants are a convenience sample. They do not stand in for domain experts, or highly specific users, or people with particular access needs. Guerrilla testing tells you whether a flow is fundamentally broken, not whether it is the best fit for your exact audience. Maze

Hallway Testing

Hallway testing is the indoor version of guerrilla. Instead of leaving the building, you stop a colleague, visitor, or in-office customer who has not seen the work and ask them to do a task while you watch.

It costs nothing and reveals obvious problems quickly, but coworkers are also unusually good at trying to guess the "right" answer. Use hallway tests to catch glaring issues like broken entry points, unreadable labels, or completely invisible buttons. Do not use them as your only signal for whether a product concept is good.

A simple rule: if someone looks confused or starts asking you questions within the first 10 seconds, something in the setup is off. You do not need a 30-minute session to confirm that.

Support Ticket Mining

Every support inbox and review feed is a research dataset that most teams barely touch. I have done "ticket mining" projects that uncovered more actionable UX issues in two hours than a month of scheduled interviews.

A lightweight process:

  1. Pull the last 50-100 tickets or reviews.
  2. Make a simple spreadsheet with columns for category, frequency, and source.
  3. Tag issues and sort by frequency.
  4. Treat the top three to five recurring categories as your highest-priority UX problems for now.

These issues come from people who cared enough to say something instead of quietly leaving, which is its own kind of filter.

If you work on B2B tools, also scan G2, Capterra, and Trustpilot for your product and competitors. Negative reviews are basically "jobs-to-be-done, but angry."

Competitor Teardowns

Use three to five competitor products and document where people will likely struggle and where the product does something surprisingly nice. The patterns that show up across tools are the mental models your users are already trained on.

When you ignore those patterns, your product feels broken before your onboarding has even started. When you lean into them, you feel "intuitive," which is often just a fancy way of saying "familiar."

This exercise also gives you a benchmark: where you are already stronger than the market and where you are not. You can see examples of this teardown style in a few breakdowns on my site, abseth.com.

5-Second and First-Click Tests

For digital interfaces, 5-second tests and first-click tests are the online cousins of guerrilla testing. They are great for quick "Does this land?" checks.

In a 5-second test, you show a screen for five seconds, hide it, then ask what they remember and what they think the product does. If people cannot tell what you do or who you are for, that is useful information before you pay for fancy illustrations.

Online Community Testing

Recruiting without a panel budget is very doable, especially for consumer products.

Some reliable places:

The single strongest pattern I have seen: emailing your users with a clear subject ("5-minute study on billing page"), repeating the time commitment, and being honest about what you are testing. Vague "we would love your feedback" invites mostly get archived.

Microsoft Clarity

For behavioral analytics, Microsoft Clarity is currently the most generous free option I recommend to teams. Unlimited traffic, session recordings, and heatmaps, with extra signals such as rage clicks and dead clicks.

The practical move is to install it once on your live product or staging environment, let it run quietly in the background for a week or two, and then look at rage and dead clicks before you schedule any user sessions. You will often discover problems that would have eaten half your interview guide.

Clarity does not have surveys or interviews built in, which is fine. It is there to show you what people actually do, not what they say they do.

The Free Stack, In One Table

Here is what a realistic zero-budget stack looks like for a solo designer or tiny team:

Method Free Tool What It Covers
Session recording and heatmaps Microsoft Clarity Continuous behavioral insight
Prototype testing (live) Figma + Google Meet Moderated usability sessions
Unmoderated prototype tests Maze free tier Flows and click analytics
5-second and first-click tests Lyssna free tier Message clarity and navigation
First-click testing ValidateThat.io Layout and IA validation
Card sorting and tree testing Optimal Workshop free or Lyssna Information architecture checks
Screeners and surveys Google Forms or Tally Recruitment and post-test data
Participant recruitment Reddit, LinkedIn, Slack People, the original panel
In-app recruiting UXtweak free widget Passive collection on your site
Support ticket analysis Your inbox + spreadsheet Direct user pain points
Competitor analysis Using products directly Benchmarks and patterns
Synthesis Modern AI assistants + transcripts Theming and summaries, with you in the loop

The combo of Clarity, Google Forms, Google Meet, and you with a notebook covers the full loop from behavior to decisions. Everything else is comfort and speed.


The Methods Ladder for One Person

Once your free tools are in place, you can scale effort based on decision stakes instead of vibes.

Level 0: Passive Research

Before you go "do research," make sure three quiet systems are running:

  1. Behavior analytics using something like Clarity. Install once and peek at it before each new sprint.
  2. Support mining, where you review tickets and reviews every so often and keep a list of recurring issues.
  3. Competitor scanning, where you try key flows in other products and jot notes.

I have seen solo designers who consistently do just these three things walk into strategy meetings with more grounded opinions than teams that spin up one big formal study per year.

Level 1: The 60-Minute Research Sprint

When you need quick signal before the next design call and have about an hour, this format pulls way more weight than it should:

  • 0-10 minutes: Write the one question this session must answer. One sentence, no cleverness.
  • 10-30 minutes: Run a heuristic review of the flow against something like Nielsen's heuristics to catch obvious stuff.
  • 30-50 minutes: Grab one person from your network or a community, share your screen or a Figma link, and have them think aloud through the flow.
  • 50-60 minutes: Capture one to three findings in the format: "[User] struggled with [element] because [inferred cause], which suggests [design implication]."

This will not hold up as "official research" for a big launch, but it is more than enough to say, "We should absolutely not ship this in its current form."

Level 2: Guerrilla, Hallway, and Community Testing

When you want a bit more confidence without blowing a sprint:

Cost
$0-50
Time
About half a day
Output
A pretty clear sense of whether the concept or flow is fundamentally broken

Run five to eight short sessions in a coffee shop, through an online community, or in your hallway. Keep sessions around 10-15 minutes, lean on think-aloud, and track task completion and friction points.

Use this for lower-stakes questions with broad audiences or when you have under 24 hours for signal. If the audience is narrow or the decision is big and political, you climb to the next levels.

Level 3: Unmoderated Remote Testing

When you need more structured data and time is tight:

Cost
Free if you DIY, around $100-500 if you use panels
Time
2-3 days
Output
Behavioral data from screened participants, no live moderation needed

Unmoderated testing gives you repeatable tasks and better sampling without needing you on every call. Five users from a single persona can surface the majority of common usability issues, though complex products often benefit from more. nngroup

This is great for "Does this flow basically work?" questions. When people behave in surprising ways and you want to ask "why" on the spot, you move into moderated territory.

Level 4: Moderated Interviews and Tests

When you need depth, nuance, and good clips for your stakeholders:

Cost
Free with your own network, $300-2,000 with recruitment tools
Time
1-5 days
Output
Rich qualitative data and strong narrative material

Moderated sessions are where solo research starts looking like "proper" research. Short, task-focused, and always recorded sessions are easier to run and share.

The common trap is leading people. Questions like "What do you like about this flow?" push them toward politeness. A simple opener helps: "There are no wrong answers. We are testing the design, not you. Please say what you are thinking as you work through this."

Level 5: Minimum Viable Study

When the stakes are high and the decision is hard to undo:

Cost
$500-5,000
Time
1-2 weeks
Output
A focused, defensible study with enough rigor to stand up in leadership meetings

An MVR setup has a clear objective, a screener, a guide, 5-8 recruited participants, 30-minute sessions, and a concise findings doc with themes and evidence. Think of it as a "study skeleton" that still has a spine.

Bring in a fractional researcher or specialist when the topic is complex or political enough that you want someone else double-checking your design and analysis. This is a rare but important move.


Building Bias Checks Into Solo Work

If you do not deliberately design for bias control, your research will quietly support whatever you believed last week. Here are a few small systems that help.

Confirmation Bias

Leading Questions

Treating N=3 as a Finding

Stopping Too Early


A Solo Research Stack for 2026

When you are ready to add a bit of budget, this is how I usually see stacks evolve:

Category Free Tier First Paid Upgrade
Behavioral analytics Microsoft Clarity Hotjar Plus for surveys and widgets
Unmoderated testing Lyssna and Maze free tiers Maze Starter with panels
Prototype testing Figma + Google Meet UXtweak or similar
Recruitment Reddit, Slack, LinkedIn, network User Interviews or similar
AI synthesis Free AI tools + transcripts Dovetail or Looppanel
Repository Notion with tags Dovetail or Condens
Screeners/surveys Google Forms, Tally Typeform
AI-moderated interviews Limited free tiers Paid plans as volume grows

For most solo designers I work with, the first genuinely game-changing upgrade is a participant recruitment platform. Tools are cheap; reliable humans are not.

AI saves time on synthesis, but it is still worth scanning raw transcripts before reading the summary. The messy parts often contain the story you really need for stakeholders.


Continuous Discovery, Solo Edition

Teresa Torres's continuous discovery model was built for product trios, but there is a lightweight solo version that actually works: one conversation per week.

This does not have to be a full-blown study. A 20-minute chat with a user, customer support rep, or sales engineer where you ask about their world (not your latest design) is enough. Take one to three notes and tag them in your log. Twelve weeks of doing that quietly changes how you design.

The trap to avoid is turning every conversation into a stealth validation session for your current feature. These are for widening your view of the problem, not proving your solution is right.


When It Is Fine to Skip Research

You do not need a study for every pixel. A simple lens:

Skip research when no one will act on the findings, when the real decision was made weeks ago, or when the primary goal is political cover. Life is short.


A One-Person Research Repository

Most solo designers have "insights" scattered across docs, decks, screenshots, and their own memory. Future you will not remember where any of that is.

Three simple habits:

  1. After each study, spend 30 minutes logging the brief, method, top 3-5 insights with quotes, and tags for product area, segment, and theme.
  2. Keep insights small and falsifiable: "New users skip the settings icon because it looks decorative" beats "Users find settings confusing."
  3. Add dates and product versions and assume anything over a year old needs a quick check before it drives a big decision.

There are fancier repository setups (I break down one on abseth.com), but even this minimal approach pays off quickly.


Selling Research When You Are the Only Researcher

In low-trust environments, "research is important" is a weak pitch. "Research saves us money and embarrassment" is better.

Some lines that tend to land:

Also, write things up. A short page with findings, quotes, and a one-sentence recommendation is something people can link and argue with. An undocumented test is just a story you tell in a meeting.


The Escalation Table

Here is a quick way to match decision type to method:

Decision Type Method Time Cost
Small tweak to an existing flow Clarity review + heuristic pass 1-2 hours $0
First impression or message clarity 5-second test 2-4 hours $0
Navigation or IA question First-click test 2-4 hours $0
Early concept for a broad audience Guerrilla or community test (5-8 users) Half a day $0-20
New feature before dev Unmoderated prototype test 2-3 days $0-200
Core flow redesign Moderated test (5-8 recruited users) 3-5 days $300-1,500
Multi-segment or leadership-level call Minimum viable study (8-12 sessions) 1-2 weeks $1,000-4,000
Strategic or hard-to-undo decision Fractional researcher or consultant 2-4 weeks $3,000-15,000

You do not need research to be grand, expensive, or beautifully branded for it to be useful. You just need it to be honest about what it can answer, sized to the decision at hand, and recorded well enough that someone else can see how you got there.