Power BI for Oil and Gas Companies: Production by Well, Deferred Volume, LOE per BOE and AFE vs Actual

What should an oil and gas dashboard measure?

An energy report earns trust when every measure has one agreed definition and a documented source. We fix those definitions with your production accountant and your operations lead before a single visual is drawn — what date a volume belongs to, whether fuel and flare come out before the number is called production, which interest applies to which measure — then implement them as governed DAX measures (opens in new tab) so the same figure survives being sliced by well, lease, basin and month.

• Volumes by stream — oil, gas, NGL and water separately and as equivalent barrels on your stated conversion, at well, pad, lease, field and basin, gross and net.
• Deferred production — potential less actual by well and day, coded by reason and costed at the realized price, so downtime is a dollar figure and not an anecdote.
• Base decline and uptime — producing days against calendar days, and the decline on the existing base separated from the contribution of new wells, so growth is not mistaken for performance.
• Lease operating expense per BOE — by cost category and by lease, with fixed and variable split, water handling isolated, and the trend read against the volume base that divides it.
• Realized price and deducts — the price actually received against the benchmark, with the differential, gathering and processing deducts, severance and ad valorem tax and royalty burden each on their own line.
• Net operating income per well — net revenue less LOE, taxes and water handling, month by month, which is what identifies the wells approaching their economic limit.
• Capital against authorization — AFE versus actual by cost code and by well, including commitments and accruals, plus cycle time from spud to rig release to first production.

Why one production number is never enough

Most arguments in an operator's monthly meeting come from two people quoting different production figures and both being right. The numbers answer different questions and are produced by different machinery, so a decision needs them on one page with their definitions attached.

The modeling problems nobody warns you about

Three things break energy reports, and all three are model problems rather than visual ones. The first is allocation: most wells commingle into a battery or central facility, so well-level production is derived from facility measurement and periodic well tests. Handled badly, a new test silently restates last month for a well that never changed. We keep measured and allocated volumes as distinct measures and carry the test date alongside the allocation.

The second is restatement. Production accounting revises prior months as tickets, statements and imbalances settle, so a report that only shows the current version cannot explain why last quarter moved. We model as-reported and as-adjusted volumes over a proper star schema (opens in new tab) with a real date table behind the time intelligence functions (opens in new tab), so a restatement is visible as a restatement.

The third is volume: a historian holding tag-level readings for every well is far too granular to import whole, and daily rates computed over a partial day are simply wrong. We aggregate at the source to the grain the report needs and use incremental refresh (opens in new tab), whose "only refresh complete days" setting Microsoft's own documentation illustrates with barrels per day — the exact trap an energy model falls into otherwise. Underneath all of it sits the well identifier. API numbers get re-completed, recompleted wells get new suffixes, and operators rename leases after an acquisition; if that key is not stabilized once, history detaches from the wellbore and every per-well trend becomes fiction.

Which oil and gas systems do you connect, and how?

We connect the systems operators actually run rather than insisting on one vendor. What varies is the route in — a reporting database, a documented API, or a scheduled export — and that choice drives refresh cadence more than anything else in the build.

Where a vendor only offers exports we say so plainly and design around the cadence that gives. Field data that arrives once a day should refresh once a day; pretending otherwise just moves the disappointment to week three.

Upstream, non-operated and midstream are different models

Operators rarely sit in only one of these. A company can operate a hundred wells, hold non-operated interests in two hundred more, and own the gathering and compression in between — and each layer measures success differently and breaks on a different join.

Who sees which wells, partners and business units?

Once a company runs more than one operating area, access becomes the constraint on a single report. A foreman should see their own wells and their own downtime without browsing another district's cost, a non-operated analyst should see the statements rather than the field, and a partner-facing view should never expose another partner's position. We implement that with row-level security (opens in new tab) mapped to your existing area, district and entity hierarchy, so one report serves the whole company and adding a field is a row in a table rather than a redevelopment. Our guide to setting up row-level security in Power BI walks through how those roles are configured and tested.

How is this different from the reports in your production accounting system?

Fairly: your production accounting software is good at the accounting month. It allocates, it distributes revenue, it produces the statements, and if that is the whole question you do not need us. What it rarely does is cross the boundary into the historian on one side and the capital ledger on the other, at daily grain, for a question nobody scoped when the system was bought. Deferred volume priced at realized value needs downtime events from SCADA, a potential estimate, an allocated volume and a price deck — four systems and a definition. That join, done once and kept clean in your own Microsoft tenant as a model you own, is the deliverable.

What does it cost, and how fast?

You buy prepaid Flex hours that never expire. Starter is 20 hours ($2,500), Growth 40 hours ($4,600) and Power 80 hours ($8,000), with senior hourly work from $100. A first build covering production and volumes, downtime and deferment, LOE per BOE and AFE against actual typically fits inside one block, and most tasks ship within 3–5 business days of your approving the scope estimate.

There is no project minimum. Every request is scoped and priced in hours before work starts, hours deduct only for work you approve, and unused hours stay banked for the next drilling program. See how prepaid Flex hours work, or what a typical custom Power BI dashboard development engagement includes.

Who is this for?

Owners, CFOs, controllers, production managers and reservoir engineers at independent and privately held operators — conventional and unconventional, single-basin and multi-basin — plus non-operated working interest holders, royalty and mineral owners with enough properties to need a portfolio view, and small midstream gathering and processing businesses. The common thread is enough wells that per-well economics stop being obvious and a team too small to keep an analyst rebuilding the same workbook each month. What changes between an oily basin and a gassy one is the price deck and the water, not the model.