If you still have to wait for a mockup for a small screen update
“We just need to add one Help item to the Settings menu.” Even when proposing a small change as a PM, you may need to explain it to a designer and wait for a mockup before you can show what it will look like. It may be a simple update in your head, but it creates one more thing you need to ask someone else to do.
What if you could make that screen yourself before the meeting? Instead of describing where things go, your colleagues could click through it and give feedback.
Figma Make’s time-saving experiment suggests that PMs may be able to turn these small changes into drafts themselves. When the same tasks were assigned to 100 PMs and product designers, the PM group with access to Make reduced its cumulative completion time across the three tasks by 23%. Let’s look at the actual tasks to see where the difference appeared.
The 100 participants were assigned three tasks
Figma recruited 50 product designers and 50 PMs, then randomly assigned them to a group with access to Make or a control group that did not use AI. Both groups completed three modification tasks on the same social media screen. Across all participants, cumulative completion time for the three tasks decreased by 20% under the Make condition.
| Experimental task | PMs | Product designers |
|---|---|---|
| Switch light mode to dark mode | Significant time improvement | No significant difference |
| Add Help to the Settings menu | Significant time improvement | No significant difference |
| Create an interaction that opens a comments panel | No significant difference | Significant time improvement |
Here, “no significant difference” does not mean the effect was exactly zero. It means the published results did not establish a statistically significant time improvement.
The work that suited PMs was small-scope screen changes
The tasks where PMs saved a significant amount of time were switching to dark mode and adding a Help menu item. The same improvement was not found for the most complex task, the comments interaction.
So it makes sense to choose a first attempt with a clearly defined scope and completion criteria. For example, add one menu item or change a specific screen state, then send a clickable draft to colleagues along with the design request. The useful takeaway is that PMs can prepare changes they once explained only in words in a visible form.
However, 23% does not mean that all design work by PMs, or all time spent waiting for designers, was reduced. It is a comparison of cumulative completion time for three tasks performed by 50 PMs, and this study did not establish output quality, collaboration effects, or project launch speed.
For designers, interactions that need to be clicked are good candidates
The task where product designers showed a significant time improvement was the interaction that opens a comments panel. The same difference was not found for familiar color changes and menu edits.
You can start with behaviors that are hard to convey through static mockups alone, such as a panel opening when a button is clicked or the screen changing based on a selection. Rather than handing every simple update to Make, choose tasks that take a long time to express for each role.
Criteria for your first use
PMs should try small-scope screen changes, while designers should start with clickable interactions. This criterion is only a pattern found across three tasks, so real teams should compare it with manual work using the same completion criteria.
Try this now: a Help menu that opens when clicked
If this is your first time, start by making one small screen without a prepared design. The steps below are a practice exercise designed to check how Make can be used, not a way to reproduce Figma’s experiment.
- Check the access requirements. Sign up or log in from the Figma Make start page. Available access varies by plan and seat, and sending prompts uses AI credits.
- Open a new Make file. In the Figma file browser, select Drafts → Make in the upper right.
- Paste the request below into the AI chat and press Send. For your first exercise, request just one screen and one behavior. Clearly state what you want to make, the elements needed, the behavior, and features to exclude.
Create a simple Settings screen that colleagues can click through during a meeting. Include only three menu items: “Account,” “Notifications,” and “Help.” When “Help” is clicked, open an information panel on the right, and close it again when “Close” in the panel is clicked. Let me verify this behavior without real sign-up or connections to external services.
Your first success is not whether the screen looks good, but whether Help opens and closes again. In the interactive preview that appears after generation, click “Help” and then “Close.” If it does not open, describe one failed behavior specifically, such as: “The panel does not open when I click Help. Please fix the connection between the menu and the panel.”
When expanding this into a real screen, you can paste the Figma frame you want to modify into Make chat and limit the scope by saying, “Keep the rest of the attached screen intact and add only a Help menu.” The official file-creation guide also explains how to bring in an existing design.
Calculate the time saved after the revision is complete
The number to track in practice is not the time when the first result appears, but the total time until a draft that meets the requirements is complete. Record all time spent writing the request, waiting for generation, and fixing failed behaviors. Also log the AI credits used for each prompt. Credit consumption varies according to the model, task complexity, and the amount of context to process.
When comparing, apply the same completion criteria to manual and Make-based work. In this exercise, the standard is: “Clicking Help opens the panel, and clicking Close makes it disappear.” If a colleague revised the result again, include that rework time too.
The most practical starting point from this experiment is clear. If you are a PM, first turn a small, clearly defined change into a working draft yourself. Then check whether it also reduces preparation time in your real work.
