How the British Heart Foundation uncovered the cost of frontline guesswork.

You have just come out of a one-to-one where a manager has raised the same frustration again. The team keeps running into the same problem. They are working around it, but staff grumbling is building.

Managers are absorbing the frustration, but the organisation is still treating it as a support problem rather than a sign that the process itself is not fit for purpose.

This is for you if you are trying to work out what to do when the real problem is easy to sense on the ground, but hard to define at the organisational level before the cost becomes too big to ignore.

The British Heart Foundation (BHF) is one of the UK’s most recognisable charities. Since 1961, it has funded cardiovascular research, staffed a helpline with cardiac nurses, and run public education on heart health. By 2035, it aims to prevent 125,000 heart attacks and strokes across the UK.

To support its mission, BHF operates one of the largest charity retail networks in the world, with nearly 700 shops. Approximately 20,000 volunteers and employees staff these shops and sell around 25 million donated items each year. For many individuals affected by heart disease, these shops serve as their first point of contact with the charity after leaving the hospital. Sarah Boardman, the Head of Retail Transformation, shares stories of people walking in and saying, “You saved my life.” The red heart logo carries significant emotional weight for them.

By late 2023, BHF had already established itself as a relatively mature digital organisation. They had migrated to Microsoft Azure, implemented Workday for human resources, and were running a multi-year technology programme that included customer relationship management (CRM) and retail finance systems. Gareth Campbell-Julian had recently joined as Director of Technology Transformation, tasked with ensuring the retail operation kept up with the rest of the organisation, but the gap he found was bigger than expected.

In 2023, BHF’s retail operations generated net profits of £24.9 million, with total income for the year reaching £180.6 million. Gift Aid, which allows charities to reclaim 25% on eligible donations, is in addition to this figure, but it can only be claimed if the item is accurately recorded at the time of donation. 

Across the UK charity sector, an estimated £560 million in Gift Aid goes unclaimed every year. The main reason for this is administrative: the necessary infrastructure to consistently capture Gift Aid across all transactions and locations is lacking. Many charities recognise this, yet it remains a low priority.

BHF, in terms of store count, is larger than many private retailers - processing more donated items in a week than some commercial retailers do in sales, so when a charity of this size and resources operates without a standard process for something as fundamental as recording donations, it raises an uncomfortable question: if smaller private businesses have established systems, why doesn’t the charity? 

The charity faced a significant challenge with 800,000 items moving through 700 shops every week. Even a small and consistent lapse in recording quality can lead to substantial discrepancies over time.

BHF had no standard procedure for recording donated stock. The problem was not that no one cared; it was that no one could see it, because the signals that something was wrong were being misread.

By the time this project began, BHF was ahead of the sector in terms of digital maturity. They had established Microsoft Azure infrastructure, implemented Workday (an HR solution), and developed a multi-year technology programme under the leadership of a CTO who had previously worked at Costa Coffee. The challenges in their retail operations were not due to neglect; rather, they stemmed from the organisation's growth and the issues that often accompany such expansion.

When charities expand to meet demand, the instinct is to measure unit performance. For example, how many shops? How many volunteers? How many items were processed? These metrics confirm that the operation is scaling, but they do not consistently show what is happening across all of the organisation’s units. Every BHF shop operated as one would expect, with donations coming in, stock moving, and tills ringing. However, in BHF’s case, it wasn't apparent that all 700 shops were making different pricing and recording decisions. This inconsistency was leading to a cumulative loss of hundreds of thousands of pounds in unclaimed Gift Aid alone. Instead of serving as a warning sign, this loss went completely unnoticed.

Meanwhile, on the shop floor, it plays out differently. When staff and volunteers encounter a decision that they lack the information or authority to make confidently, they often do not escalate the issue or refuse to act. Instead, they make their best guess and express their frustrations with grumblings. These grievances are heard by team leaders and managers, who interpret them as issues related to morale, culture, or the need for more support. However, rarely is this recognised as evidence of a missing process. As a result, the problem persists, and with BHF, this meant hundreds of individuals making the same challenging guesses thousands of times each week.

There is also a deeper structural issue affecting the entire sector. Charities open shops, recruit volunteers, and expand their services to express their mission. This drive for growth is ingrained in their purpose. However, pausing to evaluate whether the operational infrastructure can support that growth or whether process quality can keep up with the expansion creates tension with the instinct to do more. Shops continue to open, donations keep coming in, and the grumblings continue. The question of whether the underlying processes are functioning effectively isn’t a priority because everything appears to be working well, and the charity’s mission demands focus on more urgent areas. 

The catalyst for change was what occurred at BHF: this was recognised as mission-critical, not only as an internal efficiency issue but also as a stewardship concern. The unclaimed Gift Aid represented money that donors contributed in good faith to support research and services. An operational problem that results in this money remaining unclaimed is not only a financial shortfall; it creates a gradual erosion of reputational trust. For a charity that relies on public trust as the foundation of its operations, this is a particularly compelling argument for change.

If donors were aware that a donated item might not generate the Gift Aid it is entitled to because there is no consistent method for recording it, many would reasonably ask: if the process doesn’t work here, where else does it not work? Public trust in charities is currently at a ten-year high, but the most common reason for loss of confidence is concern over how donations are spent. The margin for maintaining that trust is narrower than it may seem.

In September 2024, starting with short development sprints, BHF created a guided interface (a Power Apps tool) that assists volunteers in assessing and recording each donation. The app provides recommendations for pricing and channel allocation (storefront, eBay, or Depop) based on the item's brand, condition, and seasonality. Volunteers can override these suggestions, but the default recommendation is now based on informed choices.

The AI model behind these recommendations, named Arti, was trained on 1.9 billion rows of historical retail data. Michael Spivey, a retail expert in BHF's eBay operation, highlighted the problem it addresses: previously, without reliable data, shops relied on their own judgment to make channel decisions. For example, selling an item valued at £10 would be impractical if it cost £12 to list, store, and ship it. Now, the pricing model is displayed through the app before a decision is made.

The choice was made to build this tool outside the ERP (Enterprise Resource Planning) system, creating an additional layer on top of the existing infrastructure. Campbell-Julian explained that this approach was intentional to avoid creating something that becomes harder to change the more you use it, thereby hindering future updates. By January 2025, the app was live in select stores, with plans for a phased rollout across the broader network to follow.

However, to get the most out of this case study, there are outstanding questions that the case study source does not address. The first is around how the rollout was managed at the human level. The case study source does not explain how 20,000 volunteers and staff across 700 shops adapted to this new way of working or what the change management programme looked like in practice. Additionally, it is unclear whether adoption was consistent across all locations. At this scale, the gap between a system going live and being used as intended can be substantial, and it is worth bearing in mind. 

Second, and possibly more importantly, the source case study doesn’t mention the customer perspective. The app was developed to address internal challenges such as inconsistent stock recording, missed Gift Aid, and inefficient channel allocation. While these are necessary problems to solve, BHF's shops are not warehouses. For many, the shops are emotionally charged environments, places where donors drop off items connected to their loved ones. Customers cultivate personal relationships with the charity that go beyond transactions. The source does not clarify how the introduction of a screen-mediated guided workflow was designed to coexist with these emotional relationships or whether this aspect was considered during the design process. This is noteworthy because operational systems developed for internal use do not automatically cater to external relationships such as customers and clients simply because they are well-built. 

BHF is currently processing nearly 800,000 items a week through its app. The organisation projected, as of mid-2025, an additional £1 million in revenue from improved pricing and channel allocation, as well as a £500,000 increase in Gift Aid in the first year. This transformation has won awards for Best Breakthrough Data Project and Best Not-for-Profit Project at the UK Digital Technology Leaders Awards 2025.

Beyond the headline figures, the nature of the work has changed significantly. Volunteers and staff have transitioned from making unsupported individual judgments to making guided decisions that are supported by data. Additionally, the organisation can now monitor what is happening across all 700 shops simultaneously.

Three key elements contributed to the success of this project. 

The first was speed of delivery: The team had only three months to turn the initial idea into a live product, which minimised the opportunity for the organisation to second-guess the approach. They focused on developing what they called a "minimum lovable product”, which was a functional version that users could engage with quickly and improve over time, rather than spending more time on a more comprehensive solution.

Second was the architecture: Building outside the existing ERP system allowed the team to test and refine the product without disrupting current processes. Each iteration was self-contained, enabling improvements without requiring changes to the organisation as a whole.

And lastly, how the project was communicated: Staff and volunteers were not informed that the system was evaluating their judgments. Instead, they were told that they were "teaching the machine," which framed the initiative in a more positive light.

The distinction between compliance and participation is crucial, as it differentiates a system that people embrace from one they avoid, and effective change management in a volunteer-driven organisation depends on understanding this distinction.

However, there is an unanswered question worth considering. The app was developed primarily to address stock management issues. However, the risk of creating an operationally efficient system is that it may become too inward-focused, optimised for governance, reporting, and efficiency, while neglecting the experience of the person at the checkout. Although a volunteer following a pricing recommendation may be working more efficiently, whether the system allows them to be more present and engaged with the customer remains a separate issue.

BHF’s solution involved a Director of Technology Transformation, a Solutions Architect, an existing Microsoft infrastructure stack, and 1.9 billion rows of historical data for AI model training. Most organisations reading this will not have those resources, and that's perfectly fine. 

The question being raised here is not "How do we build this?" Instead, it’s: Where in your organisation are staff making the same decisions repeatedly, lacking sufficient information, and expressing frustration about it? And what is that costing you in areas you’re not currently measuring?

Your response may require different tools, but the question applies at any scale and in any phase of growth.

A deeper structural limitation exists: operational infrastructure is often prioritised only when the mission demands it. There is a special kind of obligation for charities: they are expected to grow, expand services, open more shops, and recruit more volunteers, which often overshadows the need to assess whether the underlying processes are functioning effectively. The risk is that by the time the cost becomes apparent, for example, due to a significant error, a regulatory finding, or a reputational event, it could have been accumulating unnoticed for years.

Additionally, it is worth highlighting a notable design risk. Operational systems that are created to address internal challenges, such as compliance, reporting, and efficiency, can, if not designed with care, inadvertently lead to a worse experience for frontline workers and the individuals they serve. A system that improves operational consistency but makes human interactions more transactional has sacrificed one form of value for another. If this trade-off is not considered in the design process, overlooking it can lead to reputational risks. 

The following approaches have been inferred from BHF’s work and are also typically required to achieve similar success. These approaches are transferable to charities of any size. 

BHF is already moving beyond guided decision-making towards fuller automation, including AI-generated eBay listings and broader use of Copilot Studio. The operational logic is easy to see: if the system can identify, price, describe, and route an item faster and more consistently than a person can, why would you not let it?

But that raises a harder question than this case study answers. BHF’s shops are not just about processing stock. For many donors, the shops are places bound up with grief, gratitude, memory and trust.

The next test is one that many of us are grappling with when it comes to automation - we know it can do more work, but can charities become more efficient without becoming less human? For BHF, the spark was the moment a volunteer realised they were teaching the machine. When the machine no longer needs teaching, what are you asking your teams to contribute instead?

Every week, a real ops problem from a real charity:
What they did, what it reveals, and what you can take back to your desk.
Sign up to get it in your inbox.