Menu

Pull Quote: Failure is normal in market creation, so structured experimentation helps ensure failures are small, early, and informative, allowing learning to compound faster than mistakes.


While researching market creation stories, I recently revisited the rise of mobile money in Kenya, specifically the early days of M-Pesa. Interestingly enough, the part of this market’s story that really caught my attention wasn’t the innovation itself…it was the government’s response to it.

Instead of forcing regulation on a new business model, Kenyan regulators allowed M-Pesa to launch under supervision, gradually introducing safeguards as evidence accumulated. Regulation followed the market rather than pre-empting it. This “test-and-learn” approach helped a new market take root without abandoning oversight.

That approach reminded me of a closely related idea from the business strategy world: discovery-driven planning – a framework designed to help innovators move forward even when there’s uncertainty. 

A refresher on discovery-driven planning

Discovery-driven planning is a five-step approach developed by Rita Gunther McGrath and Ian C. MacMillan to help organisations navigate uncertainty when launching new ventures. Rather than treating assumptions as facts, it treats them as hypotheses to be tested.

The core insight is simple: the true potential of a venture is discovered as it unfolds, not known in advance.

The five steps of Discovery Driven Planning are: 
1.    Bake profitability into your venture’s plan.
2.    Calculate allowable costs.
3.    Identify your assumptions.
4.    Determine if the venture still makes sense.
5.    Test assumptions at milestones.

Done well, discovery-driven planning reduces the cost of being wrong early, before uncertainty turns into irreversible loss.

What discovery-driven planning shares with “test and learn”

The test-and-learn approach operates on the same fundamental logic. It is an experimental, data-driven method that encourages organisations and governments to start small, test ideas in real conditions, measure outcomes, and adjust before scaling.

A typical test-and-learn cycle looks like this:
1.    Form a clear, testable hypothesis.
2.    Design and implement a test.
3.    Measure impact across all channels.
4.    Roll out targeted initiative (scale what works).
5.    Learn and repeat the process.

Like discovery-driven planning, test-and-learn accepts uncertainty as inevitable. The goal is not to eliminate risk entirely, but to manage it intelligently.

That said, this approach is not without risks. As the World Bank has noted, test-and-learn frameworks can be resource-intensive for regulators, not designed to scale indefinitely, and challenging to apply evenly across participants. Without sufficient oversight, they can introduce consumer risks or competitive distortions.

Which makes real-world examples especially important.

The approach in action

For anyone out there wondering if experimentation is worth the risk, let’s take it back to mobile money in Kenya. When Safaricom first proposed launching a mobile phone-based money transfer service in 2007, Kenya’s central bank faced a dilemma. The service didn’t fit into existing banking regulations, which were designed for traditional financial institutions, not telecom companies. 

Rather than blocking the launch outright, regulators took a risk and allowed the service to proceed under a limited “no-objection” framework. Basically, the Central Bank of Kenya issued a letter of “no objection” that would allow Safaricom to launch its service as long as it met basic conditions related to consumer protection, record-keeping, and anti-money-laundering safeguards. 

This letter empowered Safaricom to launch M-Pesa, granted regulators time to observe how the service functioned in practice, and gave the market room to prove its value. Within the first nine months, M-Pesa attracted one million users and rose to four million in 18 months. The success propelled Kenya into the role of poster child for creating enabling regulatory environments. 

Why this matters for market creation

What discovery-driven planning and test-and-learn approaches share is a respect for reality. Both recognise that new markets require freedom to experiment, space to fail, and mechanisms to learn quickly.

Failure isn’t an anomaly in market creation; it’s the norm. A large share of new ventures do not succeed. But controlled experimentation helps ensure that failures are small, early, and informative rather than business-ending.

For market creation, that distinction matters. The goal isn’t to avoid failure entirely, but to build systems, whether in firms or governments, that allow learning to compound faster than mistakes.

Sandy Sanchez is a senior research associate at the Clayton Christensen Institute for Disruptive Innovation, where she focuses on understanding and solving global development issues through the lens of Jobs to Be Done and innovation theories. Her current work addresses how individuals can use market-creating innovations to create sustainable prosperity in growth economies.