# Moderated Usability Testing for B2B Websites: Give Buyers a Task, Then Watch

<a id="main-content"></a>

B2B Strategy

October 28, 2026  ·  7 min read  ·  Victor Valentine Romo

On this page

1. [Define the Decision Before Choosing Pages](<#define-the-decision-before-choosing-pages>)
2. [Give Participants a Goal Without Giving Them Directions](<#give-participants-a-goal-without-giving-them-directions>)
3. [Recruit for Relevant Work and Record the Gaps](<#recruit-for-relevant-work-and-record-the-gaps>)
4. [Set the Safety Boundary Before Opening the Website](<#set-the-safety-boundary-before-opening-the-website>)
5. [Use a Moderator Guide You Can Follow Quietly](<#use-a-moderator-guide-you-can-follow-quietly>)
6. [Record the Attempt Before Explaining It](<#record-the-attempt-before-explaining-it>)
7. [Separate Assistance From Prototype Limits](<#separate-assistance-from-prototype-limits>)
8. [Review Each Task Against Its Original Question](<#review-each-task-against-its-original-question>)
9. [Sources](<#sources>)

A supplier page lists specifications. A buyer opens three documents and still cannot tell whether the product fits.

The page has information. The buyer has an unfinished task.

Watch the attempt before rewriting the headline. Give someone a credible reason to use the website. Record where they go, what they check, and what evidence they use to decide.

[GOV.UK describes moderated usability testing](<https://www.gov.uk/service-manual/user-research/using-moderated-usability-testing>) as observing people attempting specific tasks. Its guidance calls for realistic goals and instructions that leave the route open.

The study below applies that practice to a fictional supplier. Every product, requirement, session, and worksheet entry is invented. The proposed recording rules are an editorial method for this exercise.

<a id="define-the-decision-before-choosing-pages"></a>

## Define the Decision Before Choosing Pages

Consider a fictional supplier of storage cabinets for workshops. Its website has [product pages](</articles/saas-feature-pages-vs-blog/>), specification sheets, a comparison view, and an inquiry form.

The team wants to investigate three questions. Can a buyer find installation requirements? Can they compare two plausible options? Can they identify what information an inquiry requires?

Keep the research questions separate from the tasks. “Can buyers find requirements?” belongs in the study plan. The participant needs a situation that makes those requirements matter.

For this study, invent a facilities coordinator replacing a cabinet in a small workshop. Give the coordinator a doorway measurement, an available floor area, and a list of stored items. Mark every number as an exercise input. Nothing represents equipment advice or a real purchase.

Define the expected evidence privately. A successful requirements task might include finding [the installation sheet](</articles/pdf-seo-b2b-documentation/>) and identifying the fictional clearance requirement. Clicking a preferred navigation item is not the success condition.

This differs from [customer interviews for B2B content research](</articles/customer-interviews-b2b-content-research/>). An interview can [examine a previous decision](</articles/win-loss-interview-guide-b2b/>). This session observes a particular attempt on a particular website version.

<a id="give-participants-a-goal-without-giving-them-directions"></a>

## Give Participants a Goal Without Giving Them Directions

Write task cards in the participant's language. Leave page titles, [navigation labels](</articles/tree-testing-b2b-navigation/>), and button names out of the instructions.

Use these fictional cards as a starting point:

| Task | Scenario to read aloud | Evidence for the observer |
| --- | --- | --- |
| Requirements | Your workshop has the space described on this card. Find what you need to check before deciding whether a cabinet can go there. | Participant locates relevant requirements and relates them to the supplied constraints. |
| Comparison | These two cabinets are on your shortlist. Decide which you would investigate further and explain what supports that choice. | Participant [compares relevant differences](</articles/accessible-b2b-comparison-tables/>) and identifies missing information. |
| Inquiry preparation | You want the supplier to assess your situation. Find what you would need to provide. Stop before sending anything. | Participant finds [the inquiry route](</articles/accessible-b2b-lead-forms/>) and identifies its required information. |

Keep the worksheet with product names separate from navigation instructions. A fictional product identifier gives the task a target. “Open Resources, then download Installation Guide” supplies the answer.

Check whether a task contains an accidental clue. If the website calls a section “Planning your installation,” avoid repeating that exact phrase in the scenario. Describe the buyer's concern instead.

Pilot the cards with a colleague who has not prepared them. Ask what they believe the task requires. Revise ambiguous instructions before inviting participants. Keep the final wording with the study record so later observers know what prompted the behavior.

<a id="recruit-for-relevant-work-and-record-the-gaps"></a>

## Recruit for Relevant Work and Record the Gaps

GOV.UK's [planning guidance](<https://www.gov.uk/service-manual/user-research/using-moderated-usability-testing>) recommends actual or likely users. For this exercise, recruit people whose work includes evaluating workshop equipment or gathering purchasing requirements.

Do not treat every business job title as equivalent experience. Someone who approves budgets may read differently from someone who checks dimensions. State which role the current round examines.

Use a short eligibility note: relevant task experience, recency, involvement in supplier evaluation, and familiarity with this particular site. Ask only for what the study needs. Avoid collecting employer purchasing records or confidential project details.

Record which perspectives remain absent. A round with facilities coordinators does not cover procurement review, maintenance work, and final budget approval automatically. Those are separate questions for future [recruitment](</articles/b2b-research-recruitment-screener/>).

Describe the participants and tasks in the report without [turning them into the whole market](</articles/small-b2b-customer-survey-interpretation/>). Session observations support claims about what happened under these conditions. They do not establish how often every buyer encounters the same problem.

<a id="set-the-safety-boundary-before-opening-the-website"></a>

## Set the Safety Boundary Before Opening the Website

Use an isolated prototype or a test copy with outbound actions disabled. Prepare fictional measurements, product records, names, and attachments. Provide no real payment details, credentials, employer documents, or customer information.

For inquiry preparation, stop before transmission. Prefer a mock endpoint that cannot reach the supplier. Do not rely on a reminder alone when a live button can send a message.

Check links that leave the prototype. A specification link may open a real supplier page. Document which destinations are allowed. Replace an unsafe destination with a clearly identified mock before the session.

The [GOV.UK discussion of dummy data](<https://www.gov.uk/service-manual/user-research/using-moderated-usability-testing>) warns that fictional inputs may reveal fewer contextual issues than participants' own material. Keep that limitation in the findings. This exercise uses synthetic inputs to maintain a controlled boundary.

Agree on [observation and recording permissions](</articles/b2b-research-consent-recording-scope/>) before collecting session material. State who will attend, what they will receive, and how the team will handle the records. Let participants decline recording or stop the session. Use the organization's approved research process for those arrangements.

<a id="use-a-moderator-guide-you-can-follow-quietly"></a>

## Use a Moderator Guide You Can Follow Quietly

Prepare the introduction, task cards, observation fields, and stopping rules together. The following script is invented sample wording, not a quotation from a participant or a source.

**Introduction:** We are examining this website. Your task attempts help us see what needs attention. You can pause, skip a task, or stop. Please use only the fictional information supplied for this exercise.

**Permissions:** These are the people observing. This is what we propose to record. We will explain how the records are handled before you decide whether to continue.

**Working instructions:** Please describe what you are looking for as you work. I may ask about something you do. Stop before sending an inquiry or opening anything outside the agreed test.

Read one task at a time. Keep the task card available so forgetting a fictional dimension does not become an accidental memory test.

Choose neutral [prompts](</articles/prompt-libraries-b2b-teams/>) in advance. Ask what the participant expects to happen next. Ask what information supports their choice. When they make a suggestion, ask which part of the task that change would help.

Avoid prompts that point toward an answer. “Have you tried the comparison button?” changes the task attempt. If you give that direction, record it as assistance.

Use explicit stop conditions. Stop if the participant withdraws, real information appears, or an action could contact a live supplier. Pause when a technical failure prevents observation. Explain the boundary before resuming.

<a id="record-the-attempt-before-explaining-it"></a>

## Record the Attempt Before Explaining It

Give the note-taker a record that separates actions from explanations. Use a session identifier rather than a person's name in the working analysis.

For each task, capture these fields:

| Field | What to record |
| --- | --- |
| Conditions | Session identifier, website version, device, starting page, and task card version |
| Sequence | Pages opened, documents inspected, choices made, and relevant [timestamps](</articles/event-time-processing-time-business-reports/>) |
| Participant statement | Exact wording when verified, otherwise an explicitly marked paraphrase |
| Assistance | Moderator prompt, time given, and behavior immediately afterward |
| Outcome | Independent completion, completion after assistance, incomplete attempt, or interruption |
| Open question | What the record cannot yet explain |

In a fictional session, participant P01 opens the cabinet dimensions sheet, returns to the product page, and [searches for a clearance requirement](</articles/internal-site-search-analysis-b2b/>). After several returns, the moderator identifies another document. P01 then finds the requirement.

Record completion after assistance. Also retain the earlier independent attempt. A single “completed” checkbox would hide the navigation question that made this session useful.

[Keep interpretation in a separate field](</articles/b2b-usability-findings-triage/>). “Requirements were difficult to locate from this starting page” is a hypothesis supported by the sequence. “Buyers dislike technical documents” reaches beyond the observation.

<a id="separate-assistance-from-prototype-limits"></a>

## Separate Assistance From Prototype Limits

Agree on assistance categories before reviewing sessions. In this proposed method, an ordinary request to keep describing thoughts does not supply a solution. A navigation hint does. Record the wording so reviewers can challenge that classification.

If the participant asks where to click, first ask what they would try without the moderator. Give them room to attempt it. If continued difficulty stops further learning, offer a bounded hint and mark the transition.

Keep a separate status for a broken test environment. Suppose P02 selects a comparison control that has no implementation in the prototype. The record should identify the missing behavior and the build version.

Do not convert that event into evidence that the participant could not use a completed comparison feature. You have not observed the feature. You can still record their expectation and ask what they need to compare.

Likewise, a deliberate stop before inquiry submission is the planned endpoint. Define completion as identifying the required information and reaching the agreed boundary. A stopped live submission is not a completed sales inquiry.

Preserve interrupted tasks in the report. A connection failure, withdrawn permission, or moderator intervention changes what the session can establish. An empty outcome field leaves that distinction invisible.

<a id="review-each-task-against-its-original-question"></a>

## Review Each Task Against Its Original Question

At the end of the fictional round, build a small task summary. Keep independent attempts, assisted attempts, and technical interruptions visible in separate columns. Include which sessions attempted each task.

For requirements, inspect whether participants found the relevant evidence and interpreted it against the supplied measurements. Finding a document and understanding its application are different observations.

For comparison, record the dimensions participants used and the information they could not find. Choosing an option does not prove the choice met the fictional requirement. Ask them to identify their evidence before ending the task.

For inquiry preparation, compare the information participants expected to provide with what the prototype requested. Record uncertainty without collecting real contact details.

Keep broader [conversion rate optimization](</articles/conversion-rate-optimization-b2b-seo/>) questions in a separate measurement plan. These sessions identify observed obstacles. They do not measure a conversion change or prove that a proposed revision will improve business results.

Use the resulting questions when reviewing [B2B landing page structure](</articles/landing-page-seo-b2b-conversion/>). Specify the next change narrowly, such as making installation requirements available from the product page. Test the revised path against the same underlying task.

Keep the task visible. Let the attempt supply the evidence.

<a id="sources"></a>

## Sources

- [GOV.UK Service Manual: Using moderated usability testing](<https://www.gov.uk/service-manual/user-research/using-moderated-usability-testing>). Task design, participant relevance, session guidance, and synthetic-data limitations. Checked 2026.10.28.
