Who should you recruit for website research?
Recruit research participants around a clear question. Set useful criteria, design a screening survey and describe the limits of what your findings show.
Article language: English

In this article
A friend looks at the website and says it looks great. A colleague finds the contact page straight away because she knows where you put it. A third person suggests a bigger logo. Those reactions can start a useful conversation, but they tell you little about how a new customer chooses a product. Goodwill is welcome. It rarely comes with the right recruitment criteria.
Invite people whose experience can help answer a specific research question. There is no universal “ordinary user” with one predictable way of working. Someone buying their first desk organiser, an employee who orders equipment regularly and a designer who knows the catalogue professionally may need the same information, yet read it very differently.
Start recruitment with the question. The invitation, short screening survey and session dates come later. Doing it in reverse feels efficient until the calendar is full and you discover that none of the participants makes the decision you wanted to observe.
Who do you need for this task?
Imagine a fictional shop selling equipment for a home workspace. The team wants to see how a first-time buyer compares the dimensions of an organiser with the space available on a desk. That question points to several relevant things: a recent or planned purchase, a role in choosing the item and a need to judge whether it will fit. Age alone, or an interest in design, cannot supply that information.
Prepare three groups of criteria. The first establishes whether a person's experience fits the question at all. The second helps you include differences that matter to the task. The third covers the conditions for taking part: device, preferred way to communicate and support needs. Keeping them apart makes it less likely that an open slot in the calendar becomes the main reason for inviting someone.
In this example, a core criterion might be direct involvement in choosing equipment and a real need to check its dimensions. You might also include people with different purchasing experience and ways of using the service. Do not make a long list of traits simply to make a table look representative. Each one should relate to the question or to the conditions for participation.
GOV.UK's recruitment guidance suggests that criteria can include a person's experience, circumstances and ways of accessing a service. These are categories to consider, not a ready-made recipe for every shop. Both the criteria and the number of participants depend on the purpose and the method.
Record who this round will not include, and why. If you are studying someone's first encounter with a catalogue, a regular customer might belong in a separate group. That does not make their needs less important. It means they can help answer a different question. A clear boundary is more honest than presenting every observation as the voice of all customers.
For B2B services, look at each person's role in the decision. The person finding a supplier, the person placing an order and the person approving the scope are not necessarily the same. If you are studying whether preparation requirements are understood, the experience of whoever writes the brief may matter more than their job title. “Manager” says little on its own about the task they perform.
Record these decisions in the website project brief. They help shape the scenario and make the results easier to interpret later. If the research itself is meant to reveal different roles, choose a method suited to that uncertainty. Do not pretend a precise customer profile exists simply because the document already has a heading for one. For more context, see the article on a website project brief (in Polish).
Separate knowledge of the industry from familiarity with the website. An expert may know the dimensions of common products and not need information a newcomer is looking for. A new visitor may be perfectly comfortable online but unfamiliar with the names of parts. Those are different kinds of knowledge. Recording them helps distinguish a terminology problem from an interaction problem.
A short screener should establish relevant experience
A screening survey, often called a screener, should ask about specific behaviours and situations. “Are you interested in interior furnishings?” is easy to answer yes to. “Have you recently compared equipment for your own workspace, and what part did you play in choosing it?” gives you more context for deciding whether the person fits.
Choose a suitable time frame for the product and question. A recent purchase matters if you need someone to remember particular events. A planned choice matters if you are observing a real decision over time. Do not impose the same period on every study just because the previous survey used it.
Do not reveal in the screener the detail the research is meant to investigate. If you want to know whether people find the dimensions of a base, a survey full of questions about base dimensions may already have drawn their attention there. Establish the relevant experience through a broader description of the situation, without teaching the solution to the task.
When asking about someone's role in a decision, allow more than one answer and room to explain. A person may have searched for a product for somebody else, advised on the choice or placed the order themselves. That information can help you prepare the right task. It does not have to exclude them automatically if their experience fits the aim of this round.
Offer an honest “I don't remember” or “not applicable” where it makes sense. Forcing someone to select from unsuitable answers produces data that only looks complete. Later, the team is left wondering why a person classified as an experienced buyer has never performed the task in question.
Collect only what you need to assess eligibility and organise participation. You do not need a real order number or a full purchase history to establish experience of comparing dimensions. If additional information is needed, explain its purpose and decide how to handle it before recruiting. The survey should not accidentally become a second research project.
Plan a brief follow-up when answers appear inconsistent. Someone may say they have not bought equipment, then describe choosing it for an employer. That does not necessarily mean they were unreliable. Perhaps the question suggested only a personal purchase. Check what the answer means; if the problem lies in the survey, revise it before recruiting further and keep a record of the version.

Try the screener against a few example profiles. Does it recognise someone with the right experience? Does it exclude a person merely because they use a different name for the activity? Can someone describe the support they need to take part? This is a test of the screening tool itself, separate from the later website study.
Where you invite people affects who you meet
Your customer list may help you reach people who know the product, but it is less suited to a question about a first encounter with the brand. A post in a professional group may attract experts. An invitation sent only through social media reaches people present on that particular channel. Where you look is not a neutral administrative detail.
Digital.gov's recruitment guide lists people currently trying to use a product, those who have tried it recently, and people who have used something similar and are likely to use the product in question. It also suggests different ways to reach them. Match the channel to the experience you need instead of posting an identical invitation everywhere and treating the result as a uniform sample.
Explain what participation involves: the type of activity, approximate duration, how you will meet and any materials involved. Do not promise a “short conversation” if you plan a longer set of tasks with screen recording. People need enough information to decide whether they want, and are able, to take part.
Make clear that it is not a sales conversation. Do not use the session to press for a purchase or present more packages. A participant needs to be free to describe difficulties and poor fit. A team that starts closing a sale halfway through changes the conditions of the conversation and how the result can be interpreted.
Ask about a convenient way to take part, the device and any support needed. Do not infer these from a single label. A screen reader user may need their own configuration; someone joining on a phone may need the prototype shared differently. The article on business website accessibility (in Polish) describes different ways of completing a task, which are worth considering when you organise the session too.
Do not automatically exclude someone because they cannot use the research tool you selected. First check whether that tool is necessary and whether you can prepare a suitable way for them to participate. Otherwise, the study may describe mainly people who can use the team's setup, rather than the service's audience. The article on research with assistive technology users (in Polish) explores how to prepare those conditions while respecting the participant's usual way of working.
If you offer thanks or payment for someone's time, explain the conditions clearly. Do not make it depend on a positive view of the project or completing the task without an error. Difficulty is part of the material you invited the person to help you understand. It should not become a reason for them to feel obliged to demonstrate success.
A conversation with Aeneia Design about checking a website can start with the customer's decision and the kind of experience needed to explore it. That helps define the scope before you look for the first available dates.
Keep the criteria meaningful after screening
Prepare a simple organisational list showing the conditions met, the date and the agreed way of participating. It does not have to repeat every answer from the survey. The facilitator should know why a person was invited and what context can be used in the task.
If someone withdraws, do not replace them with any available person without checking the criteria. A replacement should fit the role that matters to the study. If you cannot find one, record the change in scope. A missing group does not reappear because another friend was added to the calendar.
During a moderated prototype test, you can briefly confirm the context at the start. Do not turn it into an interrogation about survey answers. People describe their experience in different words. If a significant mismatch emerges, record it and limit the conclusions accordingly rather than pretending the criterion was met. For a closer look, see the article on moderated usability testing (in Polish).
When comparing two prototypes, account for criteria that matter to both conditions. Differences in participants' experience can make the solutions themselves harder to compare. Show in the report who saw which version, without disclosing unnecessary personal information. The article on comparing prototypes (in Polish) offers further guidance.
Choose the number of participants for the purpose, the range of needs and the kind of conclusion you want to draw. A small qualitative study can reveal specific obstacles. It does not automatically give you a representative percentage for the whole group. Many conversations with people whose experience does not fit will not repair the wrong recruitment question either.
A five-second test (in Polish) still needs the right audience, even when contact with the material is brief. It is easy to confuse a short task with no need to recruit. Who sees the offer affects what their first reaction means.
Also record how you reached each participant and their relationship with the brand. A research-panel member, a regular customer and someone responding to a public invitation may have similar purchasing experience but a different context for the session. You do not need to turn that into another reason for exclusion. The information helps you describe the sample honestly and plan how to add a missing perspective later.
After the round, review the criteria: which helped distinguish experiences, which were missing and which had no connection to the question. Do not change them retrospectively to tidy up an inconvenient result. Record what you learned for the next recruitment round alongside the limitations of the current one.
Friends can still help you check whether the prototype link works, whether a question is understandable and whether the session can be run. Name that work for what it is. In the report, distinguish this trial run from material collected with people selected for the research question. A kind compliment remains nice to hear. It stops pretending to be evidence about customers.











