esc
  • WorkflowsPages
  • Case studiesPages
  • Machine Learning FoundationsLearn
  • AI EngineeringLearn
  • AI Guardrails & Red TeamingLearn
  • UI/UXLearn
  • Design SystemsLearn
  • Production System DesignLearn
  • Book a callContact
  • Download ResumeContact
Let’s talk

I’ll build the agent for it.

Pilot: Test a Design System on One Real Team Before Full Rollout

Learn how a small, time-boxed trial shows what a design system gets wrong, what to fix first, and whether to expand it.

In short

A pilot is a small, time-boxed trial of a design system on one real team before full rollout, so mistakes affect only a few people and can be fixed first.

12 min readFree

What You Will Learn

A design system is a shared set of parts and rules that product teams use to build their screens. Before a team hands that system to every group in a company, it often tries it out first. That small trial is called a pilot. In this lesson you will learn why teams run a pilot, how to choose one, what to measure, and how to use the results to decide what happens next.

You will also follow one worked example, a shopping app that tests its new system on a single checkout flow. Read the key words below first, because they come back often.

Key words

Pilot
A small trial in which one real team or product uses something new before everyone else does.
Time-boxed
Having a fixed start date and a fixed end date, so the trial cannot drift on forever.
Design system
A shared set of reusable parts, such as buttons and form fields, plus the rules for using them.
Component
One reusable part of a screen, such as a button, a text box or a menu.
Error state
How a part looks and behaves when something goes wrong, such as a required field left empty.
Adoption
How many people actually use the system in their daily work, not just how many have heard of it.
Baseline
A measurement taken before the pilot starts, so you can later show what changed.
Consistency
The same part looks and works the same way everywhere it appears.
Rollout
Giving the system to the rest of the company, team by team or all at once.

Why Start With a Pilot?

A full rollout is costly to undo. If the system has a hole, dozens of teams may build around it at the same time. A pilot lets you find those holes while only a few people are affected. It is a way to learn before the system spreads.

  • Gaps show up early

    A pilot exposes missing parts, unclear instructions and awkward steps while the fixes are still cheap to make.

  • Learning costs little

    One team and one feature cost far less time than fixing a broken system across five or ten teams.

  • Evidence and advocates

    Figr recommends showing real-world use cases and a live demo to win over designers and developers. The people who took part often become supporters.

Pilots are also useful because they make the case for the system in plain numbers. One source describes a pilot as an opportunity to convince designers and developers that a system-based way of working is more efficient.

Keep one idea in mind. The pilot is not a demonstration that the system is finished. Its purpose is to find what is not finished yet.

What a pilot usually exposes

  • Missing components

    Teams often hit a part the system does not have on their first day. If there is no official version, they build their own and may never come back to the system.

  • Unclear documentation

    A pilot team will quickly show where the written instructions leave out a state, a size or an example they need.

  • Adoption friction

    Friction means anything that makes the system slower than doing the work another way. One source notes that teams who tried a system but did not adopt it often named missing components as the most common reason, with friction as a separate reason.

How Do You Choose and Scope the Pilot?

Scope means the exact edge of the trial: which product, which screens and which people are included. A good pilot is contained enough to manage, but typical enough that its results apply to other work. Choose it with the three tests below.

  • A willing team

    Pick a team that wants to help, will give honest feedback and will test real interactions. A reluctant team will produce weak evidence.

  • A contained but representative area

    Choose one product area, such as one flow, one feature or one settings page. It should be small enough to finish in the time you have, yet use the parts most other teams also use.

  • Parts that people use often

    Pick a feature that relies on common components, such as buttons, inputs and menus. Results from frequently used parts are seen by more people and tell you more.

Avoid choosing the most complex product you have. A huge pilot will bury the team in its own problems, and the results will be hard to read. A tiny, unusual pilot has the opposite problem, because the results may not match anyone else's work.

Tip

Write down what is out of scope, too. A short line such as 'the admin area is not part of this pilot' stops the trial from quietly growing.

Team size matters as well. Sources on pilots advise against a group that is too small to show real use and against one that is too big to coordinate. For a design system, one designer and two developers, as in our example, is a practical core group.

Setting success measures

Before the pilot begins, decide how you will judge it. Measures should be specific and checked against a target. The list below gives common options. You do not need all of them.

  • Adoption rate

    The share of the pilot's screens or code that comes from the system's shared parts, rather than custom work.

  • Design time saved

    How much less time a designer spends on a screen when using ready-made components, compared with building from scratch.

  • Development time saved

    How much less time developers spend building the same screen, compared with the usual method.

  • Consistency issues found

    How many places in the product where a part looks or behaves differently from the system's standard version.

  • Team opinion

    A short survey asking pilot members how likely they are to recommend the system to other teams.

Tip

Measure a baseline before the pilot starts. Without a starting number, you cannot prove that time was saved. Record how long a similar screen took before, and how many inconsistencies the product already had.

Fixing the timeline

A pilot needs a fixed end date. Without one, it becomes normal work and nobody can say whether it succeeded. Your timeline might be shorter or longer, but it must be decided at the start. Plan check-ins along the way, for example weekly, so problems are found early rather than at the end.

Tip

Agree on the end date and the decision meeting before you begin. This keeps the result honest, because the team will not move the goal posts once the trial is over.

Worked Example: A Shopping App Checkout Pilot

A team is building a design system for a shopping app. Instead of rolling it out to the whole app, the team runs a pilot on one area: the checkout flow, where people enter their address and payment details. The team has one designer and two developers. The pilot is set for six weeks. Here is how they work through it.

  1. 1

    Pick the scope

    The team chooses the checkout flow only. They write down that the product catalogue, account pages and search are out of scope.

  2. 2

    Set the team and the dates

    One designer and two developers are named as the core group. The six-week end date and a final review meeting are booked before work starts.

  3. 3

    Record the baseline and the measures

    The team checks how long a similar screen took to design and build before the system. They choose measures: design time, developer time, the number of inconsistent parts, and a short team survey.

  4. 4

    Build the checkout with the system

    The designer assembles the checkout screens from the system's buttons and form fields. The developers build the same screens with the matching shared parts instead of custom code.

  5. 5

    Record what breaks

    During the build, the team notices that the button and the form field have no error states. A form field needs a clear look for an invalid entry, such as a wrong card number. The team logs both gaps as the top priority, along with a note that the instructions do not explain them.

  6. 6

    Fix, measure, and decide

    The team adds the missing error states to the system and updates the instructions. At the end of six weeks it compares the measured time savings with the baseline. The result gives the team solid evidence to show managers, who then agree to a wider rollout to the rest of the app.

Tip

Notice what made the result convincing. The team did not guess how much time was saved. It measured the time against a starting number, and it wrote down the failures it found, not only the wins.

How Do You Run the Pilot and Use the Results?

Running a pilot is less about building and more about listening and recording. The team should use the same steps again and again during the trial, so that the final review rests on evidence rather than memory.

  1. 1

    Collect feedback

    Ask the designers and developers in the pilot what felt slow, confusing or missing. Gather feedback before, during and after the trial, so you can see how opinions change. Use both short surveys and quick conversations.

  2. 2

    Record what broke or was missing

    Keep one shared list of every gap. Note the component, the screen, who found it and how serious it is. Missing states, unclear instructions and extra steps all belong on the list.

  3. 3

    Fix the system

    Change the system itself, not just the pilot screens. If the team patches the checkout with custom code, the next team will hit the same gap. Update the shared part, then update the written guidance.

  4. 4

    Decide what comes next

    Compare the results with your success measures and with the target you set at the start. Then choose one of three outcomes: expand, adjust or stop. Write the decision and the reasons down.

Expand, adjust, or stop

The final review is a decision, not a celebration. It should follow the criteria you set at the start. Each outcome means something different for the next step.

  • Expand

    Choose this when the pilot met its targets and the gaps it found are fixed. The next team, or the rest of the company, can start the rollout with the improved system.

  • Adjust

    Choose this when the idea works but parts of it do not. You might add missing components, rewrite instructions or change the timeline, then run a second, smaller pilot.

  • Stop

    Choose this when the measures show the system does not help the work, or the cost of fixing it is too high. Stopping is a valid result, as long as the lessons are shared.

Tip

Share the write-up with everyone, including what failed. Sources on adoption recommend showing quick wins and measurable impact.

Mistakes to avoid

  • A pilot with no end date

    If it never finishes, it never gives a clear answer. Set the date and keep to it.

  • No baseline

    Without a starting measure, any claim about saved time is only an opinion.

  • Hiding the gaps

    If the team only reports wins, the next rollout will repeat the same mistakes. Record the failures too.

  • Treating adoption as a rule to enforce

    Some sources recommend at least one forcing function, such as a lint rule or CI check, to keep the system in use. Offer help and listen to the reasons people give for avoiding it.

Check Yourself

Try these three short scenarios. Each one needs the ideas from this lesson to answer.

Quick check

Your company has built a new design system with 40 components and wants to launch it across five product teams next week. A colleague suggests trying it on one small feature with one team first. What is the main benefit of that, and what would you measure?

Quick check

A team wants its pilot to cover the whole app, with six designers and eight developers across four products, and to keep running until everyone feels confident. Which change would most improve the pilot?

Quick check

After a four-week pilot, the team found that a date field has no error state and that two teams said the instructions were unclear. The lead wants to roll out to every team now, because the pilot was mostly a success. What is the best next step?

Key Takeaways

A pilot is a small, fixed-length trial with one willing team and one contained product area. Its job is to show what is missing, unclear or slow before the whole company depends on the system. Choose the scope with care, measure against a baseline, fix the shared parts rather than patching the pilot, and then decide to expand, adjust or stop. Done well, the pilot gives you both the fixes and the evidence needed for a successful rollout.

Sources

How is this lesson?

Last updated on 17 November 2026

Learn with me

Every lesson is free. I add new ones as I learn. Have a question, or want help with your own work? Book a short call.