Every field development plan carries an optimism nobody wrote down. Recovery factors drift up during approval, schedules drift left, and contingency is trimmed to protect the headline number. None of this is dishonest. It is what happens when a forecast is produced by a team whose next five years depend on the project going ahead.
The cheapest correction is to ask a different question. Not "what does our model say?" but "what did comparable projects actually deliver, against what they promised at sanction?" Psychologists call this the outside view, or reference-class forecasting. In field development it means benchmarking the plan before the capital is committed, not reviewing the project after it has gone wrong.
Seven families, three levels
The framework I use groups roughly forty indicators into seven families, each anchored to published benchmarks:
- Subsurface and reserves maturation: recovery factor, reserves conversion, how forecasts held up after first production.
- Project delivery and cycle time: discovery to sanction, sanction to first oil or gas, schedule slip.
- Capital efficiency: development cost per barrel of oil equivalent, cost growth from sanction.
- Operating performance: uptime, lifting cost, production efficiency.
- Portfolio and value: value created against the plan, reserves replacement.
- Emissions and sustainability: emissions intensity, flaring, decommissioning exposure.
- The benchmarking process itself: data quality, comparability, and how the result was used.
Each family works at three levels: a few headline numbers a board will read, the drivers behind them, and the diagnostics an engineer needs to explain a gap. A board does not need forty KPIs. It needs four, and confidence that the other thirty-six were checked.
Three mistakes that make benchmarks useless
Comparing projects at different stages of life
A recovery factor at sanction and a recovery factor after ten years of production are different quantities. Every record needs its maturity stage as a mandatory field, or the comparison quietly flatters the youngest projects.
Mixing currencies and years
Cost benchmarks without an escalation basis and a currency year are noise. Store the raw number and the normalised one side by side, with the basis written down.
Hiding the confidence
Some records are audited, some are press releases. A benchmark that does not carry a confidence flag per record invites the reader to treat both the same way, and to trust the outlier that supports the plan.
The single most useful column in a benchmark database is "actual versus plan". It is also the one most often missing.
Where this is going
I am working with colleagues in the Society of Petroleum Engineers on common standards for subsurface and FDP benchmarking, and building a comparator dataset from public sources, keyed by field and year, with the plan and the outcome recorded side by side. A short series of technical papers will follow.
For an operator or investor, the practical offer is simple: before the final investment decision, put the plan next to its reference class and see where it sits. If it is in the top decile on every measure at once, that is not a great project. It is a forecast that needs another look.