How Can You Tell Whether a Task Success Rate Is Good Enough?
Determining whether a task success rate is good enough requires more than looking at a percentage. First, define the task, what outcome counts as success, and the goal the evaluation should support. Then set a threshold based on the task’s importance, your own past performance, and results from comparable tests. Here, we use tree testing of website information architecture as an example. It can help determine whether users can find content, but its ratings should not be applied directly to all tasks.
Start by defining success as an outcome you can assess
Different goals call for different metrics. A 2024 NN/g article on content strategy notes that marketing content can be evaluated through traffic, engagement, and conversions, while product content can be evaluated through retention, usage, task time, success rate, and satisfaction. Content driving conversions and users finding information are therefore two different kinds of success and should not be treated as interchangeable.
In tree testing, for example, at least one correct location must be designated for each task before testing begins. Success means that the user selects a correct location as their final answer, not merely that they click on it at some point.
- A task can be stated concretely: find the library’s opening hours and directions.
- Identify acceptable answer locations in advance so the tool can score results consistently.
- If a user reaches a correct location but then goes back and selects an incorrect answer, the task still counts as a failure.
Use comparable benchmarks rather than a single passing score
In 2024, NN/g cited a review of 98 tree tests in which the median task success rate was 62%, with an interquartile range of 37% to 83%. These figures provide context for understanding this type of test, not acceptance criteria that every product should adopt.
The article emphasizes that your own previous test data provides a better benchmark, but task importance also matters. Success rates for critical or revenue-generating tasks should ideally exceed 90%. Even if a 67% success rate for finding library hours and directions falls within the suggested “good” rating range, the task may still deserve priority for improvement.
- Compare similar tasks: when information sits at different levels in the hierarchy, percentages alone cannot establish which design is better.
- Compare the same task: test it across different navigation structures or label versions to avoid mistaking task differences for differences between versions.
Tree testing deliberately removes search, visual cues, and content cues to assess the navigation structure in isolation, so its success rates are typically lower than those in standard quantitative usability testing. Success rates may improve in the complete design, provided its other elements are well designed. Poor search or menu design, however, may make results worse.
After reaching the threshold, check how smoothly users complete the task
A high success rate does not mean the process is easy. Along with completion time, examine directness—that is, whether users can find the answer without going back and trying other categories. If users succeed but take a long time, the task may still be difficult to complete.
To identify potential improvements, examine first clicks and the locations users select as their final answers. When success rates or directness are low, first clicks can indicate where users initially expected to find the information. If users click correctly at first but frequently choose an incorrect final answer, lower-level categories may overlap. Check whether the category labels are easy to distinguish.
Is the task success rate good enough: perspectives from across the web
If two navigation versions have different success rates, does that establish which is better?
Not from the percentages alone. When comparing information architectures, you can have different participants perform the same tasks in each version and compare success rates, completion times, directness, and paths. Quantitative comparisons require adequate samples. Tree-testing tools generally do not automatically determine whether differences are statistically significant; appropriate statistical tests must be selected based on the study design and metrics.
Can a success rate measured under one set of interface conditions represent all usage scenarios?
It cannot be directly generalized to other interaction conditions. In a study accepted to CHI 2026, Claude Sonnet 4.5 performed 60 everyday tasks, achieving a success rate of 78.3% under default conditions, 41.7% with keyboard-only interaction, and 28.3% when the window was zoomed to 150%. These results apply to a specific agent and set of tasks. The study tested assistive technology usage modes, not broken markup.
Does users’ recognition of interface icons prove that task success rates have improved?
No. Recognizing an icon’s function and completing a task are different measured outcomes. NN/g’s hamburger menu study found that most participants could recognize icons that followed familiar styling and placement conventions. However, the study measured only recognition, not task time or success rate, so it cannot establish whether either of those measures has improved over time. Past usability problems also involved more than icon recognition.
References
- Tree Testing Part 2: Interpreting the Resultswww.nngroup.com · 2024-01-19T17:00:00+0000 — Discusses: If two navigation versions have different success rates, does that establish which is better?
- Content Strategy vs. UX Writingwww.nngroup.com · 2024-07-12T17:00:00+0000 — Source used in the article above.
- The Text-Only Version Of Your Website Strips Out The Wrong Layerwww.searchenginejournal.com · 2026-09-25T19:00:09+00:00 — Discusses: Can a success rate measured under one set of interface conditions represent all usage scenarios?
- Tree Testing: Evaluate Menu Labels and Categorieswww.nngroup.com · 2023-08-06T18:00:00+0000 — Source used in the article above.
- The Hamburger-Menu Icon Today: Is it Recognizable?www.nngroup.com · 2025-06-13T17:00:00+0000 — Discusses: Does users’ recognition of interface icons prove that task success rates have improved?


Join the discussion 0 comments
What are you working through? Share a question or experience and let’s find the next step.
No comments yet. Start the conversation.