Greetings from Innovation Roundtable in Paris: Sharing foresight failures and tips

At the recent Innovation Roundtable event in Paris, my colleague Panu Kause and I hosted two discussions with leaders working across strategy, innovation, foresight, risk, and transformation:

Discussion 1: How does decision-making fail under high uncertainty?

Discussion 2: How do organizations make better future-oriented decisions in the age of AI?

Very few participants talked about a lack of information. Nobody said they needed more reports. Nobody complained about having too few signals.

Most organizations represented around the tables were already investing in strategic foresight, trend scouting, futures intelligence, innovation, strategy stress-testing, market monitoring, or risk analysis. Most understood foresight, and many already had systems in place. Most were also experimenting with AI. Some had dedicated foresight teams. Others had sophisticated innovation and strategy functions carrying out foresight work.

And yet a similar frustration surfaced again and again: important insights were being generated, but too often they were not influencing decisions.

The challenge has changed

For a long time, the central challenge in foresight was scanning for novelties. How do we spot emerging change early enough? How do we identify weak signals before they become obvious trends? How do we expand our understanding of what might happen next?

Those questions still matter, but in many organizations, they are no longer the primary bottleneck. Today, information is abundant. AI can summarize thousands of articles in minutes. Weak signals and phenomena can be monitored continuously. Research that once required weeks of effort can now be produced in hours. When information becomes easier to generate, attention becomes more valuable. And attention is a finite resource because it requires time, interest, and focus.

Many participants described a constant tension between monitoring broadly enough to spot important developments and maintaining enough focus to avoid drowning in noise. Track everything, and teams become overwhelmed. Track too little, and important developments emerge outside your field of view. Neither extreme is particularly useful.

The organizations making progress seem less concerned with collecting more information and more concerned with developing better ways to interpret, prioritize, and act on it. That may sound like a subtle distinction. In practice, it changes everything.

Why does foresight struggle to influence decisions?

One of the most interesting aspects of the discussion was that many failures occurred long after signals had already been identified. The challenge was not that organizations were blind. The challenge was what happened next.

Consider how most organizations operate: leadership teams are surrounded by immediate pressures. Quarterly targets. Customer demands. Operational issues. Budget discussions. Regulatory requirements. All of these deserve attention.

But they also create a powerful gravitational pull toward the present. Emerging developments rarely arrive with the same urgency. A weak signal does not usually demand a decision by Friday. A potential accelerating trend does not automatically appear on next week’s executive agenda. Budget constraints do. As a result, future-oriented insights often struggle to compete with today’s realities simply because they lack immediacy or a direct connection to sales, finance, and budgets.

Many participants described situations where important signals had been identified years before a major shift occurred. The information existed. The implications had been discussed. In some cases, reports had even been circulated. Yet little changed. Looking back, the problem was not visibility. The problem was integration. The insight never became part of the decision-making process because something more urgent was demanding immediate attention.

When measurement creates the wrong story

Organizations naturally rely on indicators to make sense of complex environments. Metrics help us reduce uncertainty, or at least the feeling of it. They help us track progress and communicate priorities. But metrics also shape what we see (or don’t see).

When organizations choose indicators that reflect existing assumptions rather than emerging realities, they can become highly confident in conclusions that are increasingly disconnected from what is actually happening. The dashboards look healthy. The reporting appears robust. The numbers move in the desired direction, or at least in a direction that can be measured with trusted indicators. This creates a feeling of safety and control. Meanwhile, the environment is changing in ways that the organization has not yet learned to measure.

This is one reason strategic foresight cannot be treated solely as a data exercise. Use foresight to understand what to measure; use indicators to understand what new changes are happening. The most important shifts are not always visible through the indicators we already have. Sometimes foresight requires us to question whether we are measuring the right things in the first place.

The hidden bias toward incremental futures

One idea generated particularly rich discussion in Paris: focus on disruptions once the usual foresight activities are up and running to spot incremental changes.

Many organizations unintentionally build systems that are exceptionally good at supporting incremental innovation. Processes are designed for efficiency. Governance structures are designed for predictability. Funding models are designed to reduce risk. Performance systems are designed around measurable outcomes. These are not flaws, but reasonable responses to the demands of running an organization focused on strengthening its existing position. The challenge is that the same structures that help organizations optimize today’s business can make it harder to explore fundamentally different futures.

What participants raised was a slowly growing resistance to disruption. More precisely, it was a lack of space for it. Future possibilities that sit close to existing assumptions can usually find support. Ideas that challenge core assumptions often struggle to gain traction because they do not fit comfortably within existing evaluation criteria and expectations.

Over time, this creates a subtle but powerful bias. Organizations can become increasingly capable of improving what already exists while becoming less capable of imagining the disruptions that might come next.

The AI paradox

Many participants shared examples of how AI is already improving horizon scanning, research, synthesis, and reporting. The productivity gains are real and the opportunities are significant, but the conversations also revealed an interesting paradox:

AI can help organizations process more information than ever before, creating a feeling of “we know it all.” If that is the case, why would we need additional foresight efforts? Yet that does not automatically mean organizations understand more or are better prepared for change.

Most AI systems are designed to identify patterns in existing data. This is incredibly valuable. However, many weak signals do not initially look like patterns. Weak signals are the seeds of patterns that are just beginning to form, or patterns that may soon fade away. You can often spot the interesting ones by observing the emotional reactions people have when reading weak signal descriptions. They emerge at the edges, challenge prevailing narratives, and may be dismissed as noise long before they are recognized as meaningful.

This raises an important question for organizations embracing AI for foresight: How do we ensure that efficiency does not come at the expense of curiosity and sensitivity to weak signals?

Several participants described using AI not to replace human judgment but to create more capacity for it. The goal was not fully automated foresight. The goal was to give experts more time to explore, interpret, challenge assumptions, and engage in collective sensemaking. That distinction feels increasingly important because the future is not simply a data problem. It remains a human opportunity.

What organizations that succeed under uncertainty do differently

The second discussion we had in Paris focused on organizations that appear particularly effective at navigating uncertainty. Their approaches varied, and a lot of interesting stories were shared. Some of the recurring themes included:

• Active foresight portfolio management: Successful organizations don’t just look at probable futures. They manage a diverse portfolio of activities exploring preferable, plausible, and even preposterous scenarios, backed by a culture and leadership that encourage looking beyond immediate operations and leveraging foresight methods that extend their perspective further into the future.

• Search for the seeds of the next business when things are going well: Periods of strong performance and peak growth are precisely the moments when organizations should invest in identifying future growth opportunities. Many organizations wait until growth slows before searching for the next growth engine. The more future-oriented organizations represented at the roundtable were doing the opposite.

• Turn business models “upside down” by reframing core competencies toward positive societal outcomes: A couple of participants described deliberately exploring what could disrupt their own business models. Naturally, they did not want dystopian scenarios to happen, but rather to use them as inspiration for creating more desirable futures to strive for.

• Strategic landscape awareness and ahead-of-the-game positioning: Rebalance resource allocation from reactive (fixing today’s problems) to proactive (positioning for tomorrow’s opportunities). Map value chains to identify where profits will shift in three to five years and stake early claims.

• Strategy as an iterative process: Rather than treating strategy as an annual planning exercise, these organizations approached it as an ongoing learning process. They experiment and test early to smooth transitions between growth cycles and avoid late-stage, panicked transformations.

• Trust-based, phased AI integration: AI is most useful when it helps organizations become more agile, rather than simply recreating existing processes. In the context of AI, success relies on building internal trust through controlled pilot projects before scaling broadly and ensuring alignment with strategic objectives.

Rather than predicting a single future, successful organizations manage a diverse portfolio of activities exploring probable, plausible, preferable, and even preposterous scenarios to maximize their adaptability. Crucially, they invest in identifying future growth opportunities during periods of peak performance rather than waiting for momentum to slow. Leaders proactively position themselves for tomorrow by shifting resources from reactive problem-solving toward understanding where value chains and profits are likely to shift in three to five years.

Finally, organizations build future agility by reframing their core business models toward positive societal outcomes, scaling AI through trust-building pilot projects, and utilizing foresight platforms such as FIBRES to support a continuous foresight practice.

Foresight becomes powerful when it becomes a system

Reflecting on the discussions afterwards, I kept returning to a simple observation: many organizations are already doing foresight activities. They run workshops, build trend reports, create scenarios, and monitor emerging developments.

The real differentiator increasingly seems to be whether those activities exist as isolated efforts or as part of a connected, balanced system.

• Can someone trace a strategic assumption back to the evidence behind it?

• Can, for example, innovation teams build on intelligence generated elsewhere in the organization?

• Can executives understand why a trend matters and how it connects to future decisions?

• Can you speak up about inconvenient changes?

• Can knowledge accumulate over time rather than disappearing into presentations and shared drives?

Foresight requires shared language, participation across functions, continuity, and institutional memory. This is one reason we are seeing growing interest in dedicated foresight infrastructures and platforms. Tools such as FIBRES are helping organizations build these capabilities by creating a shared space where intelligence can evolve continuously rather than being recreated project by project. The value is not simply faster research or better reporting. It is the creation of stronger connections between insight, collaboration, and decision-making.

And ultimately, that is where foresight creates value. Not when a trend is identified or when a report is published, but when a better decision gets made.

The future may be uncertain, but our ability to build it and prepare for it is something we can actively design. If this resonates with you, book a time to explore how continuous foresight can become part of your organization’s decision-making.