The 5 Biggest Mistakes in Mining OPEX Models

Common pitfalls that undermine cost model accuracy — and the practical steps to build models your CFO can trust.

Operating cost models are the financial backbone of every mining operation. They inform budgets, drive investment decisions, and underpin feasibility studies. Yet in our experience reviewing and building OPEX models across PGM, iron ore, copper, coal, and manganese operations, we consistently encounter the same five mistakes that erode model credibility and lead to poor decisions.

1. Blanket Escalation Instead of Driver-Based Costing

The most common shortcut in mining OPEX models is applying a single inflation rate across all cost categories. We see models where labour, diesel, electricity, explosives, and maintenance consumables are all escalated at the same 6% annual rate. This fundamentally misrepresents cost behaviour. Electricity tariffs in South Africa have escalated at rates far exceeding general inflation for over a decade. Labour costs are driven by union negotiations and are often step-changes rather than smooth curves. Diesel is linked to global oil prices and exchange rates.

The fix: Build driver-based cost models where each cost line is linked to its underlying operational and economic drivers. This takes more effort upfront but produces a model that actually reflects reality and responds accurately to scenario changes.

2. Ignoring the Relationship Between Production and Cost

Many OPEX models treat costs as either entirely fixed or entirely variable. In reality, most mining costs are semi-variable. A haul truck fleet has a fixed maintenance schedule, but actual maintenance cost varies with utilisation. Labour is contractually fixed, but overtime is driven by production demands.

The fix: Classify costs into fixed, variable, and semi-variable categories. Link variable costs to production volume or activity levels. For semi-variable costs, define the fixed base and the variable component separately.

3. Building for the Model Builder, Not the User

We regularly encounter OPEX models that are technically impressive but practically useless. They contain hundreds of linked tabs, nested IF statements ten levels deep, and formatting so dense that only the original builder can navigate them. When that person leaves, the model becomes a liability.

The fix: Design models for the next user, not yourself. Use clear naming conventions, separate inputs from calculations from outputs, include documentation sheets, and limit complexity to what is genuinely necessary. A model that nobody trusts or understands is worse than no model at all.

4. No Scenario Capability

In a sector where commodity prices can swing 30% in a quarter, exchange rates are volatile, and regulatory costs are escalating, a model that can only show one future is inadequate. Yet many OPEX models we review have no built-in mechanism for running alternative scenarios.

Mining Indaba 2026 highlighted that investment assessment frameworks now incorporate extended time horizons of 10–15 years with multi-dimensional risk analysis — making scenario capability essential, not optional.

The fix: Build scenario switches into the model from the start. Use a dedicated assumptions page where key variables can be toggled between scenarios. This adds minimal complexity but transforms the model into a genuine decision-support tool.

5. No Reconciliation to Actuals

A model that lives in isolation from actual performance data is a forecast that never learns. We see OPEX models built during a feasibility study and never updated with actual costs or variances. Over time, the gap between the model and reality widens, and confidence erodes.

The fix: Build variance tracking into the model. Include an actuals column alongside budget and forecast. Review variances monthly and use them to refine assumptions. A model that is regularly reconciled to actuals is a model that improves over time.

---

The common thread across all five mistakes is the same: the model was built to produce a number, not to support a decision. The best OPEX models are not the most complex — they are the ones that are clear, trusted, maintained, and actually used.

---

163 Solutions is an official Anaplan implementation partner specialising in the mining sector. We help mining companies migrate from Excel to Anaplan without losing the business logic that matters. Get in touch — roelof@163solutions.co.za or dean@163solutions.co.za