Skip to case study
All work Miguel Clavel Senior Product Designer
Monthly Testing Review

UX Leadership · Research Operations · Experimentation

Turning isolated tests into shared learning

Different teams were running valuable experiments, but the lessons often stayed inside each product vertical. I led a monthly testing practice that brought A/B results, user feedback, design decisions, and next steps into one conversation.

My role: UX/UI Designer and Testing Program Lead

Optimizely · UserTesting.com · Figma · Gemini · Claude

Monthly review page comparing CTA-label experiments and documenting each result, details, test page, and next step. Cover of the September 2024 Monthly Testing Review with key takeaways and the beginning of the A/B test-results section. Close preview of the Monthly Testing Review header, key takeaways, and A/B-test-results section.
The Monthly Testing Review made positive, negative, mixed, and inconclusive findings visible across teams.

At a glance

A communication system built around evidence

The deliverable was a monthly review, but the real design work was the system behind it: deciding what to test, improving the research, connecting quantitative and qualitative evidence, and helping teams act on what we learned.

{{s.title}}

{{s.body}}

The challenge

We were running tests. We were not learning together.

Each vertical focused on its own goals, pages, and experiments. That independence helped teams move quickly, but it also created a visibility gap. People did not always know what other groups had tested, which patterns were appearing across products, or whether an idea had already produced positive, negative, or inconclusive evidence somewhere else.

The problem was not a lack of activity. It was a lack of shared context.

{{p.title}}

{{p.body}}

My thinking

I reframed the problem before designing the solution

My first instinct could have been to improve individual experiments. Instead, I stepped back and looked at the path the learning took after a test ended. That revealed a larger opportunity.

I was not trying to create another report. I was trying to create a habit of shared learning.

Three moves took the work from “run better tests” to “learn as one organization.”

  1. {{sn.label}}
    2.

  1. {{st.label}}

    {{st.body}}

    2.

The system

From a test idea to a decision the team could use

The monthly recap was the visible output. The work began much earlier, with a clear question and a research plan.

  1. {{f.n}}

    {{f.title}}

    {{f.body}}

    2.

Research approach

Numbers told us what changed. People helped us understand why.

{{m.tag}}

{{m.title}}

Best for
{{m.bestFor}}
What we reviewed
{{m.reviewed}}
Important limitation
{{m.limit}}

When possible, the strongest recommendation connected behavioral evidence with the reason a design helped, or confused, the user.

Script quality

A better session starts with a better question

I wrote and reviewed the testing scripts, then refined them before launch. The goal was not to lead people toward our preferred design. It was to create tasks that felt natural, left room for honest reactions, and helped us distinguish a usability problem from a personal preference.

  • {{c.title}}

    {{c.body}}

    •

AI note

Gemini helped us pressure-test early script drafts by identifying leading wording, ambiguous instructions, and missing follow-up questions. I reviewed every suggestion and made the final research decision.

Making the learning visible

The recap turned scattered findings into a shared conversation

I created a consistent monthly format so the audience could quickly understand what was tested, what happened, what the evidence meant, and what should happen next. We shared the review across teams and discussed it in a meeting, giving each vertical access to insights that might otherwise have stayed local.

  1. {{a.title}}

    {{a.body}}

    2.

The format made uncertainty visible instead of hiding it.

Inside the review

{{g.alt}}
{{g.caption}}

{{liveMessage}}

Representative evidence

What the team could learn from individual tests

The value of the review came from presenting the evidence honestly. A positive signal could lead to a broader test. A negative result could stop an unhelpful direction. An inconclusive result could expose a measurement problem or identify the need for more data.

{{e.method}}

{{e.title}}

{{e.status}}

Observed evidence
{{e.evidence}}
Interpretation
{{e.interpretation}}
Next step
{{e.next}}

Honesty note. These labels preserve the original review’s result language. They should not be rewritten as definitive wins.

AI-assisted, human-led

We used AI to improve the test before asking people to take it

Gemini and Claude gave us a fast, low-risk way to challenge an early idea. We could ask whether a task sounded leading, whether a design created an obvious point of confusion, or what alternative explanation we might be missing. This helped us improve the material before spending participant time or launching a live experiment.

{{col.title}}

  • {{it}}
    •

AI was a critique partner, not a research participant.

Impact

The biggest change was not one winning variation. It was a better learning loop.

Individual tests produced useful product evidence. The monthly practice changed how that evidence moved through the organization.

  • {{im.title}}

    {{im.body}}

    •

What was not measured. Program-level conversion or revenue lift was not measured as part of this case study. The documented outcome is a more visible, consistent, and collaborative testing practice.

Reflection

What I learned, and what I would improve next

{{l.title}}

{{l.body}}

What I would improve

  1. {{imp.title}}

    {{imp.body}}

    2.

The takeaway

I did not just help the company run tests. I helped the company learn from them.

By connecting experimentation, user research, design critique, synthesis, and communication, I created a clearer path from isolated evidence to shared product decisions.

Return to the beginning

Role: UX/UI Designer and Testing Program Lead

Methods: A/B testing · User research · Research operations · Cross-team facilitation