INTRODUCTION

Tolling Solution

Tolling is the process of converting raw materials into finished products through third-party processors under contract. The solution covers the full workflow: contract setup, allocation of lots, production orders, and balance tracking.

Together with the product manager, I led the 0→1 process, covering weekly client interviews, prototyping and testing, field research, product design, development handover, and on-site UAT.

OVERVIEW

What is Tolling?

Tolling is a contract-based process where a company sends raw materials to a processor to be converted into finished products. The ownership of the material stays with the company, while the processor is compensated through a fee or a portion of the output.

The process is highly complex: it involves managing contracts, transport, processing times, output ratios, fees, penalties, and even pricing risks. Every step has to match up, otherwise mistakes can lead to losses.

There was no existing software that could cover the entire workflow. Most companies were still managing it with Excel and emails, which was messy and error-prone. We were able to turn these complicated rules into a clear and usable system that makes the process more efficient and transparent.

EARLY RESEARCH

Setting our preliminary research goals.

Our team traveled to London for several days of intensive on-site interviews. Through direct conversations with stakeholders, we mapped the end-to-end tolling workflow for the first time and gained a holistic view of the project.

Through our early research, we learned that the tolling workflow starts when the mine shares assay results, which trigger the creation of a lot in the system. Each lot must then be assigned a Quotational Period (QP) under strict rules before it can be allocated to a roaster. Once allocated, the company places production orders for MoOx or FeMo, and the roaster processes the material according to agreed timelines and fees. Finally, balances are tracked to reconcile inputs, deductions, and finished outputs against contractual obligations.

PROBLEM

The people we engaged with were operators from the traditional commodities industry. Their current workflows rely heavily on Excel spreadsheets for record-keeping and even fax machines for communication.

Compared to the users we typically design for, they are more conservative, less open to innovation, and prefer familiar tools over new digital solutions.

Initial Ideation

Tolling in metals is far more complex than a standard supply chain process. Unlike commodities with uniform specifications, metal lots are distinguished by grades rather than fixed standards. Roasters evaluate incoming lots by grade, and each grade comes with its own tolling fee, packaging requirement, incoterm option, and conversion timeline. These variables stack together, making cost and scheduling highly sensitive to even small changes.

What makes it more intricate is the practice of blending. Lower-grade lots can be combined with higher-grade ones to reach an intermediate quality, which reduces conversion costs compared to processing poor-quality lots separately. This blending is a paper-based adjustment that directly impacts fees and contractual obligations.

In practice, grades are not static. Lots can be re-negotiated between producer and roaster. For example, a grade C lot may be elevated to grade B if both sides agree on adjusted terms. This fluidity means that tolling systems must handle not only contractual rules like M+5 QP declarations, obligations, and balances, but also real-world dynamics like grade negotiations, blending strategies, and variable conversion times.

Principles

Replicate real-world workflows as closely as possible, so the tool feels familiar, minimizes learning costs, and avoids adding operational burden.

Final Design

Contracts

Contracts are the fastest way to understand how the entire tolling process works. Each contract typically comes with a wide range of clauses, often varying from case to case. To bring structure, we analyzed existing agreements, identified the most common and universal terms, and designed the system to support them directly while also allowing optional inputs for customization.

In current practice, teams rely on large Excel spreadsheets, cutting them into sections and building formulas to simulate allocation outcomes. These ad-hoc models help them decide which contract and roaster will maximize value, but the process is manual, error-prone, and hard to maintain.

To support this, we extended the platform’s existing contract workflows by introducing a contract split feature. Once a parent contract is defined, users can quickly generate child contracts that reflect specific terms or timeframes. Alongside this, we integrated Excel reading capabilities, enabling users to analyze and reference contract data directly within the system.

Final Design

We divided the feature into two main views. The first is a balance view, modeled after users’ Excel habits but built on a structured data model for clarity. The second is an allocation view that integrates balances through the concept of batches and combines delivery with allocation. A calculator at the bottom displays floating results to reduce manual work and better reflect real supply chain logic.

These decisions not only matched user expectations but also significantly reduced development effort. Early testing showed strong acceptance from users, while the system’s architecture remains scalable for future enhancements.

Final Design

Final Design

After the MVP release, we added reconciliation and formula features based on back office needs. In tolling, balances are not static—they must be recalculated against assays, contracts, and operational outcomes. The client’s data scientists had built models to predict these outcomes, but users still needed to reconcile them with reality.

The initial request was an Excel-like type editor, but this was too heavy for our team to implement. Instead, we broke it into two parts: a modal input tool that transforms all provided equations into configurable fields, and a formula preview so users can quickly verify if their inputs are correct.

Outcome

The tolling platform not only digitized a highly manual process but also delivered measurable business impact across efficiency, cost, and compliance.

More

Tolling is not only complex for our customers, but also for colleagues who had never seen the process before. To make it tangible, we created a role-playing game called Tollerama, where teams acted as producers, processors, and consumers negotiating under real market rules.

In just a few rounds, colleagues experienced the same trade-offs and tensions our users face every day. This playful approach helped the entire company understand tolling better, align on its value, and even enjoy the learning process.

Learn more →