Five Failures from the Frontier

A blog by Lil Patuck, a member of the Frontier Tech Hub

How would you pick up a box? Credit: Greg McKinney

If you'd like to read about why we embrace failure as one of our most valuable assets, you can start with the first piece in this series: Why Failure Matters.

In this piece, we’re diving into five specific stories from our portfolio of pilots to demonstrate what failure really looks like in practice. These span the tiny tweaks and big-time losses: the moments that lead to better solutions, sustainable pathways to scale and more profound impact for those who need them most. 

If every pilot was successful and we did not make any early exits, it would mean that we were not pushing the technology frontier after all. That is just not what we are about.
— Bojan

How would you pick up a box?

Tracking aid with GeoSeals, Ethiopia

After two years of research, planning, and testing, a health worker in Harar, Ethiopia, picked up a heavy box and walked through a warehouse doorway. The edge of the doorway held a freshly installed 'geoseal hub' drilled into the concrete just minutes before. Inside, wires connected a chipboard and sensor to a local SIM card, built by hand by design engineer Ruby Hill 6,000 miles away on the Isle of Wight.

On the other side of the doorway, the man set the box down and turned to a confused group crowded around a laptop who had fallen silent.

"We didn't expect him to hold the box like that," said one person eventually, while another laughed and scrolled through unchanged code.

The hub was designed to detect a signal from a sticker as the box passed through, triggering a report that life-saving malnutrition supplements had arrived at this rural health center hundreds of kilometers from Addis Ababa—a journey where aid can easily be lost, stolen or delayed.

Ruby had spent weeks testing sensors on her own doorways back on the Isle of Wight, varying box contents, sensor heights, and antenna placements. But in all their tests, one thing they hadn't accounted for was how a person might actually carry the box. In their tests, boxes were carried with two hands in front. Here, the box was hoisted onto the man's shoulder and carried through on its side, hiding it from the hub's view.

The pivot

This single moment of real-world testing revealed a critical blind spot, so the team could redesign the sensor placement and expand the detection field to account for natural human carrying behaviours. It became an insight into how to make the system work for the people who would use it every day. Sometimes the most valuable test costs nothing more than watching someone walk through a door.

Read more about this story here, or listen to the podcast episode.


A firefighter attempts to put out a forest fire in Brazil Credit: ERALDO PERES / AP

The wrong user

Using drones and nano-biogel to prevent wildfires in Brazil

Not all pivots come from dramatic failures. Sometimes the most valuable learning happens when you catch a false assumption before it becomes an expensive mistake.

This pilot launched with a clear vision: develop a biodegradable gel that could be deployed via drones to combat wildfires, with firefighters as the primary users. It made intuitive sense—firefighters fight fires.

During the first sprint, the team worked with a drone specialist, who analysed the nanobiogel's properties and delivered unexpected news: the gel's characteristics limited drone dispersal to a smaller radius than anticipated. It was more suitable for containing wildfire outbreaks than tackling the large-scale wildfires already burning. The technology could work—just not in the way they'd envisioned.

The pivot

They recommended targeting Brazil's widely used crop-spraying drones and engaging directly with farmers, rather than firefighters. This led to shifting the tech use and rethinking the user. They rapidly tested this new direction, securing £10k co-funding from SEBRAE (a Brazilian startup accelerator) to demonstrate that the nanobiogel could be deployed via crop-spraying drones.

This is Lean Impact methodology in action: debunking critical assumptions early, before significant resources are committed. Rather than spending months developing a solution for firefighters only to discover the mismatch later, they validated the riskiest assumption first. The "failure" happened in a £10k sprint, not a £100k scale-up. Sometimes the best failures are the ones you catch early enough that they barely feel like failures at all—just smart course corrections on the path to impact.


The unexpected stall

Smart Mobility for Urban Accessibility in Kenya

In order to begin user testing, you need to have the product available to test. When the Smart Mobility pilot set out to test their electric third wheels for wheelchair users in Nairobi, they faced a seemingly simple challenge: getting the devices there.

Transporting the equipment via commercial flights proved far more complex than anticipated. Size restrictions, airline regulations, and the intricacies of shipping mobility equipment created cascading delays and budget overruns. What should have been straightforward logistics became a major bottleneck, consuming time and resources initially earmarked for the actual testing.

Despite these setbacks, the team successfully completed user testing, learning significantly about the feasibility of inclusive mobility devices in Nairobi's unique urban environment. But with limited time and resources remaining for Sprint 3 activities—which focused on further testing and employment model exploration—continuing down the original path would have meant rushing through critical work or abandoning valuable insights already gathered.

The pivot: Rather than force-fit the remaining activities, the team redirected their efforts toward creating something with lasting value: a living blueprint. This playbook for developing and implementing inclusive mobility solutions for wheelchair users in Nairobi captured their hard-won lessons about logistics, local context, and user needs. While they couldn't complete their original testing roadmap, they ensured future innovators wouldn't face the same expensive surprises. The project's goal of sustainability through learning-sharing was met—just not in the way they'd originally imagined. Sometimes the most valuable output isn't the product you set out to test, but the roadmap you create for those who follow.


Credit: Raimond Spekking

The country pivot (and the rest that followed)

Predicting displacement in Ukraine

Sometimes, a single pivot before a contract is even signed can reshape everything that follows. The PRDICT pilot was initially designed to test predictive displacement modelling across Mozambique, Niger, or Nigeria, using various data sources, including climate data. But then, after discussions, Ukraine was chosen instead for its potential to attract Ministerial interest.

This wasn't a small tweak. Choosing Ukraine meant sacrificing the climate data sprint in favour of satellite data and entering a conflict zone where the volume and intensity of displacement dwarfed anything the team had modelled before. Conflict data from the International NGO Safety Organisation showed more incidents in a single day in Ukraine than had been reported in weeks in Nigeria or Mozambique. The team's usual method—assessing displacement per incident—collapsed under the scale.

What followed was a cascade of smaller pivots, each one revealing something valuable: abandoning grid-based networks for road networks, as testing showed that people actually travel along roads, not abstract grids. Dropping state-sponsored movement modelling when interviews revealed it had tapered off by 2024. Reframing Google Trends from a timing predictor (when will people move?) to a destination predictor (where will they go?) when the data told a different story.

The pivot

Each adjustment forced the team to rethink established methods, generating insights that will shape future displacement modeling far beyond Ukraine. What started as a strategic decision to pursue funding became a masterclass in adaptive modeling. Sometimes the biggest learning comes not from the failure itself, but from the dozens of smaller course corrections it demands along the way.


When the company goes bust

ARVR for conflict-related sexual violence

Some failures hit suddenly and completely. The AR/VR pilot was progressing well, exploring how immersive technology could support survivors of conflict-related sexual violence through mental health interventions, courtroom familiarization, and crime scene reconstruction for the International Criminal Court (ICC). Then the implementing partner went bankrupt, halting efforts to scale the project.

It's the kind of setback that could render months of work meaningless: a partner collapse with no warning, and for many traditional programs, this could just be filed under "lessons learned" and quietly forgotten.

But the pilot had already produced something valuable: a comprehensive report exploring how AR/VR tools could integrate into ICC workflows. The research was solid. The potential applications were clear. The insights about trauma-informed immersive technology were groundbreaking.

The pivot

Rather than let the partner's bankruptcy bury the work, the team shared the report with the ICC, where it sparked meaningful conversations about integrating immersive technology into their operations. The insights contributed to growing recognition of AR/VR's value in complex legal processes, particularly for sensitive cases involving survivors of violence. While the pilot couldn't scale as originally envisioned, the knowledge didn't disappear—it became a foundation for future innovators working at the intersection of justice, technology, and trauma support. The company may have gone bust, but the learning compounded.


Fail fast. Share learning. Pivot to improve.

This is our approach on the Frontier Tech Hub, and we’re growing our portfolio to learn even more from pivots and shifts that occur when an assumption turns out to be inaccurate.

As we said in our previous piece: in an era of budget constraints and heightened scrutiny over public spending, betting big on solutions that look promising on paper is a luxury the world can't afford. History's spectacular failures serve as expensive reminders of what happens when we skip the messy, uncomfortable work of failing small and learning fast.

Look out for our final piece, where we’ll delve into the broader lessons these failures have taught us—from the unpredictability of user behaviour in real-world conditions to the unique challenges of designing for government adoption. These insights, harvested from a decade of strategic failures, continue to inform how we harness frontier technology for impact.


Frontier Tech Hub
The Frontier Technologies Hub works with UK Foreign, Commonwealth and Development Office (FCDO) staff and global partners to understand the potential for innovative tech in the development context, and then test and scale their ideas.
Next
Next

Why failure matters