First-Click Testing: The First Click Predicts Success
First-click testing measures where a person clicks first on a task. How to set it up, what to measure, how many people it needs, and when it does not fit.
How is it done?
The procedure is simple. The person sees a screenshot or prototype and receives a task framed as a scenario, for example “You want to change the shipping address of your order.” They click once, where they would look first. The team records the exact point, the time, and whether the click falls in the correct zone, which is defined in advance (Koji: First-click testing guide).
The Koji and CleverX guides make two recommendations. First, tasks should not repeat the words on the screen, because that hands the person the answer. Second, the correct zone must be defined precisely before the test: a zone that is too broad inflates the success rate (CleverX: First-click testing methods and benchmarks).
What is measured?
The Koji and CleverX guides agree on four measures:
- A heatmap of clicks, to see where attempts concentrate.
- The first-click success rate, meaning the share of clicks that land in the correct zone.
- Time to the first click. A fast, correct click suggests the option is obvious; a long delay may signal doubt.
- Open reasoning, if you ask why the person chose that spot. It is the part that helps most in understanding an error.
These guides also offer ranges for interpreting the success rate: above 80% counts as a strong result, and below 60% they recommend redesigning or not moving to development without changes. These are conventions from tool vendors, not validated thresholds; use them as orientation.
Why does the first click matter?
The most cited figure comes from Bailey and Wolfson, who analyzed first clicks on functional websites between 2006 and 2009, across about a dozen studies: 87% of users whose first click was down one of the optimal paths eventually succeeded, versus only 46% of the others. They did not use a click test to generate those findings. MeasuringU tried to replicate it with five websites, 21 tasks, and 750 participants and found a more modest effect: 78% success after a correct first click versus 70% after an incorrect one, a 13% relative increase. In a tree test (137 tasks, 1,000 participants), the gap was 81% versus 14% (MeasuringU: Do Click Tests Predict Live Site Clicks?, 2023).
The cautious conclusion: a correct first click helps people finish the task, but the size of the effect varies by type of site and type of test. Do not use it as a fixed probability.
How many people do I need?
Vendor guides give different ranges. Koji cites 15 to 30 per task for a directional test and puts the useful point for iterating on an information architecture at around 30. CleverX proposes 20 to 30 for early decisions, about 50 as the standard, and 100 or more for comparisons. These are industry conventions, not study results, so treat them as starting points. If you need firm numbers, check sample size against the question you want to answer.
When should you NOT use it?
- In multi-step flows. The method measures one click, not the full sequence. A task-based test serves better for that.
- In complex forms, where the problem is understanding fields, not finding a screen.
- When emotional response or comprehension of content is what matters.
- With tasks that repeat words from the interface, because the click becomes a text match.
- With too many tasks on one screen. Many tasks in a row create artificial scanning and fatigue.
Before choosing the method, ask which decision depends on the answer. If the decision is to change a label, the first click is enough. If it is to redesign a flow, you need more.
The method also does not replace a research plan built around the tasks people really do in context. It shows where people click, not why they trust one place over another.
Illustrative example
Illustrative example: an appliance store wants to know where people look to “change my shipping address.” It shows the same screen to 30 people with the task written as a scenario. Forty percent click “My account,” 30% click “Orders,” and only 20% click the section the team expected, “Shipping and delivery,” which has an overly technical name. Asked why, several people say “Shipping” sounds like something for the store, not for them. The team renames the section “My delivery address” and repeats the test with 30 new people. Successful first clicks rise, and the time to the first click falls.
If you want to organize navigation before testing, review navigation patterns. And to make labels clear, microcopy is the next piece to check.
Common mistakes
- Using words from the screen inside the task.
- Defining the correct zone after seeing the results.
- Running many tasks in a single session.
- Recruiting only colleagues, who know the product and do not represent users.
- Treating a success percentage as sufficient evidence without asking why.
- Concluding the design works because the first click was right, without checking the rest of the flow.
What you can do tomorrow
- Write three tasks as scenarios, without using any word from the screen.
- Draw the correct zone for each task before showing it to anyone.
- Show the screen to five people and ask why they chose each click. That tells you which label to change before running a larger test.