Measuring how much effort a task takes requires more than checking whether users ultimately succeed. First define the task and context of use, then examine success rate, path directness, and completion time together. Among the sources discussed here, the most direct evidence comes from tree testing: observing whether users can find the correct location within an information hierarchy, whether they backtrack, and how long they take. These metrics can indicate how difficult it is to find information, but should not be treated as a universal effort score for all tasks.
First, clarify what needs to be done and who will do it
A task must be specific enough to assess whether users succeed and how smoothly they get there. Nielsen Norman Group uses the example of finding where to look up tuition fees. This tests whether users can locate the information, not whether they can complete a payment. These tasks have different endpoints: finding the tuition page does not mean the entire payment task is complete.
The context of use also includes users’ familiarity. Nielsen Norman Group notes that path directness is particularly important for tasks frequently performed by novices or occasional users, who cannot rely on past experience to learn and remember where information is located. When interpreting results, describe the participants’ backgrounds and task conditions rather than generalizing one group’s performance to all users.
Consider success and directness together
Success rate measures how many users find the correct location; path directness shows whether they backtrack or switch categories along the way. Nielsen Norman Group cautions that even with a high success rate, the process may still be difficult if users repeatedly go back and try multiple categories. The following outcomes can be distinguished:
- Direct success: Reaching the correct location without backtracking or switching categories.
- Indirect success: Eventually finding the answer, but taking a wrong turn and going back to search again.
- Direct failure: Going straight to the wrong location without backtracking, believing the answer should be there.
The third outcome shows that a short path with no hesitation does not necessarily mean a good experience. In the tuition lookup example, some successful participants still had to backtrack. Reporting only their eventual success would overlook the effort involved in finding the information.
Use completion time to find clues, then observe where difficulties arise
Completion time can supplement path data. Nielsen Norman Group recommends first comparing the time taken for different tasks within the same study to identify those that take relatively longer. Even when users ultimately succeed, a longer completion time may indicate difficulty and mental effort. However, this is only a signal worth investigating: time alone cannot establish the cause, and there is no universal time threshold for passing every task.
To understand the difficulties more fully, observe what users actually do. CXL distinguishes between two methods: heuristic analysis uses UX principles to assess whether the user journey is clear, whether the content is relevant, and where friction occurs; usability testing observes real people completing tasks to identify where they get stuck. The former identifies potential problems that need verification; it cannot replace the latter. When compiling results, record the task, whether users succeeded, their backtracking paths, and their completion time. This allows judgments about user effort to be checked against observable behavior.
Measuring user effort: perspectives from across the web
What task-completion time can be used to assess pages receiving AI referrals?
Search Engine Journal recommends auditing the ten landing pages receiving the most AI referral traffic, checking whether visitors can complete their intended task within 30 seconds of arrival on each page. This is a proposed audit approach for pages receiving AI referrals and can help identify pages worth further investigation. The material does not establish 30 seconds as a passing threshold for every task, nor can that time alone determine how much effort visitors expend.
How can pages reduce friction when visitors arriving through AI referrals are ready to act?
If visitors have already decided to book, check prices, or get started before arriving, Search Engine Journal recommends moving the corresponding access points to the top of the page so they do not first have to navigate to another page or scroll past content that reintroduces the product. This recommendation assumes that visitors are ready to complete the task. Whether it actually reduces effort still needs to be observed for the specific task and context of use; it should not be assumed that all visitors arriving through AI referrals have already made a decision.
References
- AI Mode Sends A Different Visitor. Your Website Wasn’t Built For Themwww.searchenginejournal.com · 2026-06-19T09:00:22+00:00 — Discusses: What task-completion time can be used to assess pages receiving AI referrals?; How can pages reduce friction when visitors arriving through AI referrals are ready to act?
- AI for CRO research: what to automate (and what not to)cxl.com · 2026-08-20T13:22:22+00:00 — Source used in the article above.
- Tree Testing Part 2: Interpreting the Resultswww.nngroup.com · 2024-01-19T17:00:00+0000 — Source used in the article above.


