Testing With Puppets: What I Learned About Data-Driven Content Before Netflix Made It Famous

In 2013, I sat in my office at Outfit7 watching a crude puppet show we'd just filmed on the cupboard behind my desk. Talking Tom, Angela, and Ben—characters from our mobile games—brought to life using the official plush toys we'd manufactured but couldn't yet sell.

The production budget was under £200. The "studio" alternated between my office and my intern Calvin's dinner table. The production quality was laughable.

But the YouTube analytics were beautiful.

Those puppet videos went viral. More importantly, they taught me something about combining data with creative instinct that would guide my work for the next decade—across mobile games that reached a billion downloads, health platforms that scaled to 30 million users, and retail innovation labs at one of the world's largest energy companies.

This is the story of how I stumbled into data-driven content development years before it became conventional wisdom, and what I learned about making smart bets when you can't afford to be wrong.

The Problem: Expensive Bets Without Evidence

When I joined Outfit7 as Chief Operating Officer in 2013, we had a challenge many companies face: how do you expand a successful mobile game franchise into other media without betting millions on unproven concepts?

The traditional approach in entertainment was clear: hire expensive studios, produce high-quality content, launch on established platforms like Cartoon Network or Nickelodeon, and hope for the best. An animated TV series could easily cost €10 million or more. This was how the industry had always worked.

But I'd just come from HiT Entertainment (owned by Mattel), where I'd been responsible for some of the world's most beloved children's properties—Thomas the Tank Engine, Bob the Builder, Fireman Sam. I'd seen firsthand how expensive it was when you got it wrong. I'd also negotiated one of the first major Netflix deals for children's content, which gave me an early view into how streaming and digital distribution were changing everything.

More importantly, I'd spent years analyzing TV ratings and audience data. I knew that YouTube analytics gave us something traditional TV never could: real-time feedback on what actually worked.

So we tried something different.

The Methodology: Test Cheaply First

Before investing millions in animation production, we tested the concept in the cheapest way possible. My intern Calvin and I made simple puppet videos using the official Talking Tom plush toys we had sitting in the office—the ones we'd manufactured but couldn't yet sell. Some videos were filmed on the cupboard behind my desk, others on Calvin's dinner table. We uploaded them to YouTube.

The total budget was under £200. The risk was minimal. But the learning was massive.

The videos went viral. Not because of production quality—they looked exactly like what they were, homemade puppet shows—but because the characters and storytelling resonated. We could see it in the data: completion rates, engagement patterns, which jokes landed, which characters people loved, how long they watched.

This gave us confidence to move to the next phase: low-cost 3D animated shorts, about two minutes each. We launched these simultaneously in our mobile app and on YouTube, using both platforms to gather analytics. Which storylines kept people watching? What pacing worked? How did different age groups respond?

We iterated based on what we learned. Each short was an experiment that informed the next one. We were building a data-validated creative formula before committing serious money to full production.

Only then did we produce the high-production-value MyTalkingTom TV series. Here again, I applied what I'd learned: I negotiated with a talented new production company that wanted to prove themselves, securing quality animation for significantly less than half what established studios would have charged. And because we had evidence the content would work, we took a risk that seemed crazy at the time: we secured YouTube Kids' first-ever exclusive content deal and launched there before traditional television.

Everyone in the industry said you had to launch on TV first. We had data suggesting otherwise.

It worked.

Why This Approach Mattered

This wasn't just about being scrappy with limited budgets, though that helped. It was about a fundamental shift in how to make decisions under uncertainty.

The traditional approach assumed you needed complete creative vision and conviction before you built anything. You imagined the perfect product, invested heavily, and hoped the market agreed with your vision.

The data-driven approach says: test your assumptions as cheaply as possible, let evidence guide your creative decisions, then invest confidently in what you've validated.

But—and this is critical—it's not about letting data make creative decisions. Some elements will always come from experience and creative instinct, that ability to see patterns and opportunities that data alone can't reveal. The key is knowing when to trust your instinct and when to demand evidence.

As we often said at Outfit7: "Data tells you what happened. Instinct tells you why. Together, they tell you what to do next."

The Results

The MyTalkingTom mobile game launched in 2013 became the most downloaded mobile game in history, surpassing one billion downloads. The broader Talking Tom & Friends franchise became one of the most successful mobile entertainment properties globally.

The TV series succeeded because it was built on validated creative decisions. We knew what worked before we spent serious money producing it.

Outfit7 was eventually sold for over $1 billion in 2017.

I'm proud of these results, but I'm more interested in why the methodology worked—because I've since applied the same approach across completely different industries with similar success.

One Methodology, Multiple Industries

After Outfit7, I co-founded Healthily (originally Your.MD), a digital health platform. Here we applied the same test-iterate-scale approach to building OneStopHealth, a marketplace connecting users to health services.

We didn't launch with 100 partners. We started with one. We built comprehensive analytics to track the entire conversion funnel - from our app through to partner services. We tested pricing models using Facebook and Google analytics to understand what drove conversions.

Once we understood what worked for both users and partners, we expanded to five carefully selected partners, refining our integrated analytics system. Only then did we scale to over 100 partners - because we had proven the model works and had the data infrastructure to manage growth effectively.

We grew to over 30 million users. More importantly, we achieved Class IIa medical device certification, demonstrating that data-driven development and regulatory rigor aren't mutually exclusive.

Most recently, as CEO of BP's Convenience and EV Labs, I applied this methodology to retail innovation. Our team introduced the concept of a Test Store - a dedicated environment where we could trial new technologies and services, gather real customer behaviour data, and iterate before rolling out across BP's network.

One example: we developed a smart kiosk system initially designed to remotely unlock refrigerated vending machines at forecourts. But as we gathered data on how it was actually being used, we realised we'd built something more valuable - a platform for remote management of any hardware at forecourts. We pivoted the technology to enable control of battery rental systems, car washes, and other equipment. The same core technology, validated and then adapted based on what we learned.

This approach led to innovations that are still being implemented and driving clean tech adoption across BP's operations today.

The pattern is consistent: test cheaply, validate with data, iterate based on learning, invest in proven concepts.

It works in entertainment. It works in health. It works in retail. It works because it's not industry-specific - it's about how to make smart decisions when you can't afford to guess.

What Netflix Made Famous (That We Were Already Doing)

Years after we pioneered this approach at Outfit7, Netflix became famous for data-driven content development. They use viewing data to decide what to produce, test concepts before major investment, make renewal decisions based on engagement metrics, and personalise recommendations based on behaviour.

This is exactly what we were doing in 2013 with puppet shows filmed on office furniture and YouTube analytics.

The difference is scale and sophistication. Netflix has massive budgets and advanced analytics infrastructure. We had plush toys from our warehouse, Calvin's dinner table, and freely available YouTube data. But the fundamental principle was identical: use evidence to reduce risk, validate creative instincts before making expensive bets, and let the data guide - but not dictate - creative decisions.

I take no credit for inventing this approach. We were responding to constraints and opportunities in front of us. But we were early, and being early taught me lessons that continue to shape how I think about innovation and transformation.

What I've Learned About Data and Creativity

After applying this methodology across three different industries for over a decade, here are the principles I keep coming back to:

1. Start with the cheapest possible test

   Don't build the full product to test your hypothesis. Find the smallest, fastest, cheapest way to learn if you're right. Under £200 in puppet videos taught me more than a €10 million production ever could have.

   
   

2. Data is essential but not sufficient

   You need data to validate your instincts, but you need creative instinct to know what questions to ask. Data tells you what happened; instinct tells you what it means and what to do next.

   
   

3. Test in the real market, not in research

   YouTube taught us more than any focus group could. Real people making real choices in their real context is the only feedback that truly matters.

   
   

4. Iterate quickly with low-cost experiments

   Every experiment should inform the next one. Build a learning loop, not a single bet. At Healthily, going from 1 partner to 5 to 100+ wasn't just scaling - it was learning at each stage what made partnerships successful.

   
   

5. When data validates your instinct, move boldly

   The YouTube Kids exclusive deal seemed risky to others but felt obvious to us; we had evidence. Use data to give you confidence to make contrarian bets.

   
   

6. Stay flexible on the "how"

   At BP, a system built to unlock fridges became a platform for managing multiple types of forecourt hardware. The best innovations often emerge when you pay attention to how people actually use what you've built.

A Note on Being Wrong

I should mention: not everything worked. We had tests that failed. We had ideas that the data told us wouldn't work, and we were right to kill them early. That's exactly the point; better to discover something won't work with a £200 puppet test than a €10 million production.

The methodology doesn't guarantee success. It just makes failure cheaper and learning faster.

Why This Matters Now

We're in an era where companies face constant pressure to innovate but can't afford to make expensive mistakes. Whether it's AI integration, digital transformation, new product development, or market expansion - the challenge is the same: how do you move forward when you can't be certain what will work?

The answer isn't to avoid uncertainty. It's to embrace it with a methodology that lets you learn quickly and cheaply.

Test your assumptions. Use data to validate your instincts. Iterate based on what you learn. Then invest confidently in what you've proven works.

This works in content. It works in health. It works in retail. It works because it's about decision-making under uncertainty - and every industry faces that.

What's your experience? Have you found ways to combine data with creative instinct? Or seen companies struggle because they relied too heavily on one or the other?

I'd be interested in hearing your stories—drop a comment or reach out at info@radycle.com.

Follow our posts

Love to #radycle? Good news!

You can add our tags (#radycleblog) throughout your breathing to reach more depth. Why hashtag? People can use our hashtags to search through content on our blog and find the content that matters to them. So go ahead and #radycle away!