TracrFirst product hire at a diamond traceability startup, where the platform could not reliably identify the stones it was supposed to track, and the pivot that fixed it
1. The Context
Tracr was a stealth-phase startup backed by De Beers, building a blockchain platform to trace diamonds through the supply chain. I was recruited by BCG Digital Ventures as the first permanent product hire, reporting directly to the CEO. My mandate was to turn an ambiguous concept into a usable product.
The blockchain gave each diamond a digital twin: an immutable record of its origin and journey, transferable between parties as the stone moved through the supply chain, with each party able to append their own data to it. Matching each physical stone to its digital twin reliably was the problem the platform had not yet solved.
My first major task involved a project called “Sight Week Preview.” Sight Week is a periodic sales event where De Beers presents parcels of rough diamonds to approved manufacturers. As an incentive to participate in our pilot, manufacturers were given an early preview of the stones they might buy at the upcoming Sight Week. In exchange, they had to match those diamonds to their digital IDs on our platform.
“First thing I noticed is that Steven was very quick to pick up an in depth understanding and the needs of different aspects of a complex industry. He used this to then quickly drive product ideas to meet industry requirements. He was instrumental in implementing an agile approach to doing this, something that had been very much lacking in the organisation prior.”
2. The Problem
Very quickly, I discovered a critical flaw in the application’s underlying logic: traceability was being managed through a spreadsheet where the primary identifier was a diamond’s weight in carats. Because many diamonds share identical weights, stones of identical weight were indistinguishable from one another.
Manufacturers were calling to ask which stone was theirs. The system’s design left the question genuinely unanswerable. This undermined trust in the pilot and exposed a fundamental issue: reliable identification of the physical stone was the foundation traceability depended on.
Product thinking: Reliable identification of what is being tracked is the foundation of any traceability system. The identifier problem had to be solved before anything else mattered.
3. The Pivot
From a product perspective, I identified that the core issue was relying on a single measurement as the identifier. Working closely with Nick, the Head of Data Science, and his team, we explored how to build something richer. The solution combined 3D model matching with weight and other metadata to create a composite fingerprint unique to each physical stone.
This approach faced two significant practical hurdles:
- A lack of descriptive language. Unlike polished diamonds, rough stones lacked any standardised language for describing their features or orientation. Defining something as simple as “top” or “bottom” for an irregular stone required solving a problem the industry had never needed to solve before.
- Encrypted data. The industry uses Sarine machines to scan diamonds, but the resulting files were encrypted by the vendor. Even when manufacturers wanted to share their data, they were technically unable to do so.
“Steven always brought to our meetings good energy, coupled with open mindedness and critical thinking that challenged and encouraged ideation. Steven has an ability to quickly understand complex structures and problems and come up with intuitive solutions and improvements, all while taking customers needs and expectations in mind.”
4. Pragmatic Execution
Given these constraints, I prioritised speed over precision. Working closely with the data science team, we aligned on building lower-fidelity 3D models that we could control and iterate on. Lower fidelity than the Sarine scans, but data we owned and could work with immediately.
We started with a matching confidence score of approximately 50%. Through continuous iteration, we improved this to over 70%, the threshold needed to make the pilot viable for manufacturers.
This removed the need for manufacturers to manually maintain fragile links between physical stones and spreadsheet IDs, restoring trust in the pilot and enabling a scalable approach to traceability.
5. Strategic Impact
Beyond solving the immediate problem, this pivot established the foundation for treating each diamond as a “digital asset.” Strategically, this allowed De Beers to differentiate natural diamonds from lab-grown stones by anchoring each stone to a verifiable story of origin.
The platform tracked diamonds from rough stone through cutting plans, polished outputs, and grading certification, with each stage linked back to the original. A single rough diamond could yield multiple polished stones, each tracked as a child of the rough and carrying its full lineage forward.
It also opened up future opportunities, such as “digital planning,” where a specialist could map out the optimal way to cut a rough diamond and attach that plan to the digital asset before the stone was ever sent to a manufacturer.
Strategic thinking: Traceability solved a supply chain problem. But making every diamond independently identifiable solved a market problem: it gave De Beers the mechanism to distinguish natural stones from lab-grown ones at the point where that distinction was becoming commercially existential.
“[Steven is] able to contribute to or lead discussions, user testing sessions and workshops, [and] present and articulate thoughts in a clear and confident manner. I learned a lot from Steven due to his in depth knowledge of building products in a digital space and his extremely detailed documentation practices.”