Desk research: investigating without talking to users, step by step
Desk research means investigating with information that already exists: studies, internal data, and outside sources. Learn when to use it and its limits.
What is desk research?
There are two broad families of research. Primary research means collecting data yourself: interviews, usability tests, or surveys. Secondary research means reviewing what others have already found. Nielsen Norman Group defines secondary research as gathering prior research findings and other relevant information for a new project, and notes that it is also known as desk research or, in academic contexts, a literature review.
The name refers to staying at your desk instead of going into the field. It is not an exotic technique. It is the logical first step in almost any design project.
What sources should you review?
Nielsen Norman Group separates sources into two groups:
| Type | Examples |
|---|---|
| Internal | Past studies, journey maps, service blueprints, consultant deliverables, workshop results, customer service and search logs |
| External | Academic journals, the ACM Digital Library, university libraries, Google Scholar, open data repositories, white papers from reputable organizations |
Nielsen Norman Group suggests that internal material ideally lives in a research repository. In practice, your first sources are usually the ones your company already has: support tickets, app store reviews, product analytics, or satisfaction surveys.
Why does it matter to a designer?
Because it saves you weeks of questions that already have answers, and it lets you walk into interviews with concrete hypotheses. A designer who reviews support tickets and earlier studies first asks better questions in the first session.
It also helps you scope the problem. Nielsen Norman Group describes it as a standard first step in any rigorous research practice and says it is often cost-effective, even in casual settings: it lets you build on earlier work instead of reinventing it. With that, you can decide whether the problem deserves an investment in your own research.
How do you do it, step by step?
The guides agree on a similar process. Here is one you can follow tomorrow:
- Define a specific question. “Why do users abandon checkout?” works; “understand our users” does not.
- Choose sources. Prioritize credible, relevant, and recent ones.
- Extract data with discipline. Note the source, date, sample, and whether the data confirms, contradicts, or is silent on your question.
- Sort the findings. Koji proposes three groups: what is well established, what is contested, and what is unknown. The unknown part is what you should investigate.
- Write a short summary of one or two pages covering what you know, what you doubt, and what you want to validate, and share it with the team before primary research starts.
Nielsen Norman Group suggests a similar four-step path: choose the topic and problem statement, identify internal and external sources, collect and critically analyze the data, and summarize the findings with your team to decide what still needs investigating.
To judge a source, Nielsen Norman Group gives these cues:
- Peer-reviewed publications are generally reliable and include a section describing methods, data-collection techniques, and limitations. If it is missing, look harder.
- Be wary of blog posts that are opinion with no primary research behind them.
- In quantitative studies, look for confidence intervals and statistical significance.
- Ask whether the findings apply to your audience: a pattern validated with young adults may not hold for older users.
What are its limits?
Nielsen Norman Group warns that secondary research is not a substitute for primary research (doing both is best) and that its quality depends heavily on the quality of its sources. Koji adds four concrete limits:
- Organizations publish successes. Case studies rarely describe what failed.
- Reports arrive late. Published reports typically reflect data collected 6 to 18 months before publication, which matters most in fast-moving markets.
- They describe averages, not your users. They do not show your specific users in your product’s context.
- They tell you what happened, not why. Understanding the why in enough depth requires primary research.
So desk research tells you where to look, but it does not prove what happens with your specific users.
Example
Illustrative example: a team wants to redesign the sign-up flow of a personal finance app. Before interviewing anyone, they review six months of support tickets, read the app’s store reviews, and look for studies on sign-up forms in financial services. They find that the most repeated complaint is SMS verification, not the form itself. With that hypothesis, they prepare interviews focused on verification instead of the whole form.
Common mistakes
- Treating an opinion article as a study without checking who wrote it or what data it uses.
- Using data from another market or country without checking whether it applies.
- Stopping at reading and never writing a summary with decisions.
- Assuming desk research is enough and skipping research with people.
- Not recording the source, so nobody can verify the data months later.
If you need your own data after the desk research, user interviews and user surveys are the natural next steps. To compare products that already exist, competitive analysis goes deeper into that part. You can also use concept testing to validate ideas before building them.