A website at the real workstation: what a quiet test will not show
Observe a site at an actual workstation: choose the task, account for the environment, record actions without inventing causes and test design changes.
Article language: English

In this article
The instructions open quickly, the button has good contrast and a test participant finds the next step easily. Then the same page reaches a phone propped up at an assembly table. One hand holds a component, the other searches for the right picture, the screen dims and someone nearby asks for the kit number. The quiet desk session was not wrong. The assembly table simply did not receive an invitation.
Observation at the place of use shows the site as part of a larger task, alongside the product, packaging, notes and other people. Choose it when a design decision depends on conditions that a mock-up alone cannot reproduce credibly. It does not automatically represent every user. It shows particular journeys and helps form hypotheses for further checking.
Start with the activity that needs the site
Imagine a fictional company, Moduł. Customers assemble light display organisers using instructions on its website. The team wants to understand how they find the correct variant and return to a step after an interruption. It is not studying the entire assembly job or the participant's skill. The subject is use of information in an agreed task.
“Is the site convenient?” leaves too many possible answers. A narrower question is: how does a person check that the instructions match the kit on the table? Another concerns returning: after speaking to someone nearby, how do they recognise the last step completed? These questions lead to observable actions.
GOV.UK describes contextual research as observing activity in its everyday setting, with participants' own equipment and ordinary interruptions. Its guide distinguishes quiet observation, occasional questions and explaining each step, noting that questions affect the natural flow. The method should fit the question; not every session must be run identically.
At Moduł, first observe how the participant reaches the instructions independently. They might start with a label on the box, a saved message or a search engine. If the researcher opens the correct page for them, a stage the company wants to learn about is skipped. Define the task's starting point when preparing the visit instead of accidentally simplifying it.
Record materials the participant normally has at hand. If the variant label is in a message, they switch between it and the page. Opening the instructions on the researcher's extra laptop would remove that switching. Their own phone, normal browser and real entry point may be part of the question, not obstacles to a more convenient session.
Decide who belongs in the observed group. Someone assembling an organiser for the first time and an employee preparing a dozen kits may need different information. An experienced participant may skip explanations because they know the product. Not using a passage does not immediately mean it can be removed for everyone.
The groups may start in different places too. A newcomer looks for the kit name; an experienced person returns directly to an unusual joint. Preserve these differences when planning visits instead of forcing identical starting points. In a later comparative trial, conditions can deliberately be standardised if that comparison supports the design decision.
Product instructions for the first steps (in Polish) should clearly identify the variant and stage. Observation checks how someone recognises that scope beside the physical kit. Compare what they see on the page with the labels they actually use in their environment.
Preparation should not iron reality flat
Agree the purpose, task, place and documentation with participants and those responsible for access. Explain who will observe, what will be recorded and why, who can access the material and how it will be stored securely. Before taking notes or recording, obtain every participant’s informed consent to the agreed scope. Explain that scope to anyone joining later and obtain their informed consent before documenting their participation. If a need to capture new material arises, reconfirm consent for that scope. Before photographs or recording, also arrange the site host’s permission where needed.
If the question concerns finding the right step, the record should show that journey. There is no need to film the whole workstation without a connection to the purpose and agreed scope. A photo of the product and phone arrangement may help, but should include only material approved for capture. Customer names, notifications and other people's documents do not become necessary research data merely because they enter the frame.
Before the visit decide when observation will stop. Normal work needs, participant discomfort or departure from the agreed scope take priority over finishing the scenario. Do not create difficulty merely to obtain more interesting material. In Moduł's example, observe ordinary use of instructions without provoking unsafe situations or asking for unusual handling of equipment.
Leave the environment as needed for the task. If someone usually props their phone against the box, record that rather than immediately bringing a convenient stand. If they use a sheet with labels, examine its role. The researcher may propose a solution later, but should not remove the problem before understanding it.
A sheet of paper can be an effective tool rather than evidence that the website failed. A participant may mark their progress on it or pass a kit number to someone else. Find out what it enables. The design might support working with that note instead of insisting that every action move onto the phone screen.
Run a short organisational rehearsal with the team. Check that you can take notes without obscuring the participant's screen, read them afterward and keep equipment out of space needed for normal work. This tests research conditions. It is not yet customer observation or evidence of website usability.

Separate behaviour from the explanation that comes to mind
A participant returns to the step list four times. A note saying “does not understand the instructions” adds a cause not yet confirmed. They might be checking a component number, returning after an interruption or comparing variants. First record the action and context. Later clarify it with the participant and available material.
The GOV.UK guide to research notes recommends recording what was seen or heard rather than the researcher's interpretation. It separately describes photos and recordings. Records should allow the sequence to be revisited; agree their scope before capture, including when something new arises during the visit.
A fictional Moduł note might say: the participant puts down a component, scrolls to the heading and compares a label with the underside of the box. Only in conversation do they explain that they were unsure of the variant. This suggests a design hypothesis about identifying instructions. It does not yet prove every user needs the same highlighting.
Keep three things separate: what was observed, what the participant said and how the team interprets it. Their explanation is important evidence, but need not describe every unconscious step. If the account and observed sequence differ, leave a question for further checking. Do not choose the more convenient version merely because it makes a simpler slide.
Ask questions without suggesting an evaluation. Instead of “is this button too small?”, return to the specific action: what did the person want to do then, and how did they recognise the result? If you must interrupt the task, record when. A later time analysis should account for the pause the researcher created.
Similarly record your own assistance: pointing to a link, explaining a name or holding the phone. Success after such intervention is not the same as independent completion. Show in the report where support changed the result and what remains untested without it.
Do not choose a fix so early that you stop following the rest of the task. Returning to the start may indicate a navigation problem, but it can also be a sensible way to compare two pieces of information. Record whether the participant achieved the goal, what they used and what made work difficult. Select a solution after gathering evidence.
A new button will not fix every difficulty at a phone
Suppose one participant cannot return to the previous step because the screen timed out during an interruption. In another case the page is still open, but the heading does not show where they are. These are different problems, though both may look like renewed searching on video. The design should address the actual cause and the service's limits.
The Moduł team might test clearer variant labels and an easier return to the step. It need not rebuild the whole instruction immediately. First specify the expected effect: after an interruption, the participant should independently recognise the right place. Then select a trial that can show this.
For each proposed change keep a short chain of reasoning: observed moment, task difficulty, hypothesis and expected effect. Repeated variant comparison might lead to testing its label at every step. If the trial does not reduce uncertainty, revisit the explanation of the cause rather than declaring the new highlight successful because of its appearance.
A moderated prototype usability test (in Polish) can check the changed layout in a controlled task. A first-use cognitive walkthrough (in Polish) helps analyse in advance what the user must recognise at the next step. Both complement the visit; neither automatically recreates conditions noticed at the table.
If a task lasts several days, one observation may not show returns to the instructions. A diary study (in Polish) then answers a different question about the course of events over time. Choose a method for the missing information rather than adding more identical sessions merely because the team already knows how to run them.
Present results with the visit's limits: who participated, the task, equipment and conditions affecting the sequence. Do not present one observation as a frequency across the whole group. You can identify an important problem without inventing a percentage of customers who experience it.
A result can remain unresolved. If no return after a break occurred during the visit, absence of a problem does not confirm that stage works. Record what was actually observed and choose a separate way to check the missing situation. The visit's scope is defined by the material collected, not every question originally written into the scenario.
Further checking can include a task performed without a moderator (in Polish) if it suits the question and material. Retain the fact that the observation method differs. No researcher beside the participant does not mean you automatically know all the conditions around them.
For a conversation with Aeneia Design, bring one task, a description of its setting and the decision you want to make after research. This helps limit the visit to what is needed. The website need not win a contest for the quietest screen. It should help a person finish work with the product under the conditions that actually accompany the task.







