When innovation fails, it comes down to how decisions — not ideas — are made

This article is based on a webinar (now available on-demand) featuring Rupert Naylor, Senior Vice President for Test & Learn at Mastercard, Forrester analyst Boris Evelson and Matias Faret, who has spent the last several years working for global restaurant chains at Restaurant Brands International. The webinar unpacks findings from a recent Mastercard commissioned study with Forrester Consulting on experimentation’s role in driving innovation — especially as today’s influx of data insights and ideas, market disruption and AI transform how we do business.

• New research reveals that companies recognize the need to innovate but lack the infrastructure, processes and culture required to turn ideas into strategic growth 
• In an environment defined by disruption and AI‑driven change, experimentation enables leaders to move quickly without relying on instinct alone or making untested, high‑stakes bets 
• 80% of surveyed organizations are looking for capabilities to test initiatives on a small scale before full rollout, as small-scale, well-designed tests help organizations manage risk, build confidence in decision‑making and accelerate innovation over time, according to experts 
• Cultural resistance remains a major barrier, as employees often need clearer leadership signals to view failed tests as learning opportunities rather than failures

Today, innovation is a must. But despite heavy investment in AI, transformation initiatives and bold new strategies, most companies come up against a formidable challenge: They can’t reliably turn ideas into lucrative outcomes. A 2025 Mastercard-commissioned Forrester Consulting study, “The Experimentation Advantage: Research On Derisking Innovation,” shows that organizations overwhelmingly recognize the need to innovate — yet they’re short on the data insights, talent, processes and cultural norms required to translate their initiatives into market success.

As Forrester VP, Principal Analyst Boris Evelson puts it, the issue goes beyond ambition. “It's creating the environment on guardrails that makes taking smart risks — not just risks — possible.” With AI adoption accelerating and competitive pressures intensifying, the cost of an ill-chosen bet has never been higher.

Today, Corporate America’s obsession with “disruption” is at a saturation point. Under pressure to move fast, many organizations default to improvisation, relying on instinct or informal consensus to place bets that can cost millions. Others are too cautious, trickling out half‑measures that leave both consumers and stakeholders dissatisfied. This dual bind has left 87% of senior leaders struggling to balance innovation with risk management, according to the Forrester Consulting study.

The strongest innovators find a pragmatic middle ground. They recognize that a steadfast testing protocol ensures good ideas reach the market faster, rather than slowing their ascent. As Rupert Naylor, Senior Vice President at Mastercard, notes, leaders need “a consistent process, so you’re not changing how you evaluate ideas every time.” Employing a unified measurement methodology also enables consistent comparison across initiatives and levers — a significant asset in the AI era, where technology is constantly evolving and iterating. 

Big ideas are expensive to get wrong.  It tracks, then, that 80% of global leaders believe the ability to test initiatives on a small scale would accelerate innovation and help manage risk. These capabilities are now essential for true transformation. 

No one articulated the stakes more clearly than Matias Faret, a marketing and analytics leader for several globally recognized brands. In industries like quick service restaurants and retail — where national campaigns require significant media spending and operational adjustments — an untested idea can quickly become a multi-million-dollar misstep. As he put it, “Risks are necessary, but if you are able to take calculated risks, then you’re going to be better off than going in completely blind.”

Even small-scale tests fail when they are poorly designed. Many organizations try to test too much at once, producing results that are ambiguous or misleading. “I see tests fail because they don’t answer the question, or they try to answer 20 questions at once,” Faret said. Developing a concise hypothesis from the outset will generate more actionable results.  

Elsewhere, teams begin testing without clearly defined objectives. Without shared goals established upfront, interpretation becomes more subjective later in the process, and different stakeholders may draw disparate conclusions based on their perspectives or priorities. This variability makes it harder to reach alignment and can reduce confidence in the overall testing approach. 

Senior leaders and teams at all levels can encounter natural biases, often shaped by prior experience, established preferences or operational priorities. These factors can inadvertently influence how results are interpreted. This is why strong experimental design — with clear hypotheses, predefined success metrics and structured interpretation — is the linchpin to maintaining consistency and confidence in the process.

The cultural obstacles to experimentation are often more persistent than the technical ones. Without clear guidance from leadership, employees may interpret an unsuccessful test as a personal setback — especially in environments that emphasize bold ideas over the learnings derived from them. But as Faret noted, an inconclusive or negative test can be valuable, as it helps organizations avoid costly missteps. Leaders who recognize and celebrate avoided risk as much as positive outcomes help create a climate where teams feel comfortable testing early and often. Naylor agreed that leadership plays a central role; executives must normalize trial and error and clearly reinforce that setbacks are part of the learning process, not career‑limiting events. 

By fostering a culture that embraces learning through iteration, senior leaders also encourage openness around results. Naylor highlighted the importance of communicating findings in clear, practical terms: “Don’t present data in charts without any explanation… speak in the language of the business.” When data insights are shown alongside helpful commentary, teams are more likely to understand, trust and apply them.

While many leaders view AI as a way to fast-track decision‑making, it does not reduce the need for testing — it amplifies it. AI can surface patterns and potential opportunities at unprecedented speed, but it cannot determine which ideas merit investment without structured experimentation and human judgment. Think of it this way: AI accelerates, but experimentation validates — and together, they generate the insights that move business. 

For brands without a consistent experimentation strategy and clear leadership alignment, AI can also inadvertently reinforce existing assumptions or biases rather than improve outcomes.  Faret cautioned against magical thinking, saying “AI is not a crystal ball; the strongest outcomes happen when humans and AI work together to design and interpret experiments.” This spirit of collaboration can help testing become more reliable and strategically grounded.

As organizations scale experimentation, best practices become essential. Yet nearly seven in 10 surveyed leaders say that aligning standardized experimentation and analytics practices across their organization is one of the most difficult obstacles in pursuing innovation. Evelson explained that such practices — governance frameworks, clear measurement models and dedicated teams — don’t slow innovation down; they enable it,” creating the consistency required for innovation to be sustained rather than sporadic.

The broader lesson is that innovation is no longer a matter of inspiration alone. Ideas are abundant; follow-through is scarce. The companies positioned to outperform in the coming decade will not be those that chase every new trend or deploy AI the quickest, but those that develop a durable system for testing, learning and scaling. They will anchor their innovation bets in evidence rather than intuition. They will value tests that prevent mistakes as much as those that reveal breakthroughs. And they will recognize that, in a world defined by volatility, experimentation is the real competitive advantage. 

To see how these principles play out in practice, watch the on-demand webinar featuring industry leaders and subject‑matter experts, as they unpack how experimentation helps organizations move faster without increasing risk. You can also explore the full research to dig into the data behind the discussion and understand where organizations are still falling short. Beyond the study itself, our website brings together a broader set of perspectives — research, expert insights and real-world examples across experimentation, AI and advanced analytics — to help leaders translate learning into action, strengthen decision confidence and scale what works with greater discipline.