Walnut Coding
A UX research case study on post-trial parent decision-making for children's coding education.
About Walnut Coding
This research examines how parents decide whether to continue after a trial coding class. The work maps evidence chains, hesitation signals, and communication opportunities that help families understand learning value without relying on promotion-heavy follow-up.
Role
UX Researcher
Team
Research lead,
Business stakeholders
Tool
Interview synthesis,
Decision mapping,
Opportunity framing
Company
Walnut Education
Year
Research dossier
The trial class is the moment a parent moves from “interesting” to “I’ll pay.”
Walnut Coding teaches computational thinking through Scratch, online classes and project-based exercises. In children’s education, the trial class is the critical conversion moment — but a child enjoying it does not, by itself, open a wallet.
Basic reading, computer operation, and the ability to explain a simple project.
Cares about logical thinking, the learning path, and return on education spending.
Recommendation intention is the wrong metric for a trial class.
Parents haven’t experienced formal classes, homework feedback or long-term progress yet. So I shifted the research lens from satisfaction toward payment confidence, perceived value, perceived risk and path clarity.
“How do we make parents more satisfied or more willing to recommend?”
“Why do parents hesitate after a trial class, and what evidence do they need before they’re willing to pay?”
A high-decision-value segment, supported by internal direction and public data.
Children this age produce observable learning behavior in a Scratch trial, while parents form more complex payment judgments. Public Scratch data backs 8+ as the right boundary for Scratch-based research.
Is my child interested? Do they understand basic logic? Can they complete work independently?
Is Scratch too basic? Can it connect to Python? Does the course have long-term value?
“Child likes it → parent pays” is a myth. The real chain branches.
The trial doesn’t only need to generate interest. It has to walk the parent down a ladder of uncertainties — and every unanswered rung delays the payment.
Four insights into why “fun” doesn’t convert.
The trial class is a parent risk-reduction moment, not an interest showcase.
Beneath “Did my child enjoy it?” parents are eliminating risks: temporary excitement, a harder formal course, heavy parent support, Scratch being too basic, an over-polished trial. Fun proves interest exists — it doesn’t remove uncertainty.
The barrier to payment is not lack of interest. It is unresolved risk.
Parents aren’t buying a coding class — they’re buying an explainable investment rationale.
Many parents can’t judge whether dragging blocks meant real learning. They need a framework to justify the spend to themselves. That’s why the child’s ability to explain the project — translated by the teacher — turns output into visible learning evidence.
Parents are buying a credible explanation that their child is growing.
For ages 8–12 the real tension is entry-point legitimacy: “Is Scratch still right?”
Most of these parents already accept that coding has value. Their hesitation is the path: Scratch now, or jump to Python? If the product only says Scratch is fun, it feels too young. If it shows how Scratch builds program logic, structure and debugging before Python, it feels professional.
The key isn’t proving coding is fun — it’s proving Scratch is a reasonable entry point.
Competitors don’t just compete on features — they offer different kinds of reassurance.
Each alternative gives parents a distinct sense of security. Walnut isn’t only competing against products; it’s competing against the reassurance each option provides — and must answer each one directly.
When parents compare alternatives, they compare which option makes them feel more secure.
What parents asked for, by intensity.
Reconstructed and illustrative — these show the analytical structure and recurring patterns, not company metrics. The interpretation: post-trial conversion was blocked less by price than by unresolved uncertainty and weak learning evidence.
Scoring pain by frequency, intensity, alternative pressure and conversion impact.
The top problem isn’t “make the trial more fun.” It’s giving parents credible evidence and reducing uncertainty after it.
Six moves that trade persuasion for evidence.
Each opportunity maps back to an insight and turns the post-trial moment from a sales pitch into decision support.
Risk-Reduction Post-Trial Report
Reframe the trial summary as evidence: what the child did independently vs. with hints, concepts touched, observed risks, and what the first four formal weeks look like.
Child Explanation + Teacher Translation
In the final 5 minutes, the child explains one command and one problem; the teacher translates child language into parent-readable learning value.
Age-Specific Trial & Learning Path
Different post-trial narratives for 8–10 (interest, logic, independence) and 11–12 (variables, conditionals, Python transition).
Decision-Support Comparison Page
Don’t avoid competitors — help parents choose by situation. Advisory tone beats “we’re better than X.”
Performance-Based Enrollment
Recommend a starting level from observed behavior, not a discount — turning a sales decision into an education-planning decision.
Formal-Course Support Mechanism
Clarify roles (child / teacher / parent) so busy parents know they won’t end up teaching the course themselves.
Parent-facing learning evidence beat promotion-driven follow-up.
In the real business context, related changes to post-trial parent communication were later validated through internal experiments and showed positive directional impact on trial-to-paid conversion. The strongest pattern wasn’t more promotion — it was making the child’s learning evidence and the formal-course path clearer. To protect confidentiality, no setup, sample size, traffic allocation or exact uplift is disclosed; the learning is presented at the level of strategic direction.
Where the thesis could be wrong.
Good research names its own counter-signals. These keep the recommendations honest and the next round of validation sharp.
Three experiments to run.
Which learning evidence matters most?
Project explanation vs. teacher interpretation vs. written report vs. learning-path vs. first-four-week expectation.
Which segment benefits most from age-specific design?
Compare 8–10 vs. 11–12 vs. high-intent vs. Python-comparison vs. free-tutorial-comparison parents.
How much comparison should be shown?
Does transparent comparison raise confidence, or quietly increase decision delay?
The real goal of a trial class is to build an evidence chain.
Not just to prove coding is fun — but to prove this child can start systematic learning here, and that the parent understands why the investment is worth it.