How to read BSEE decommissioning data

The regulator publishes almost everything. Permits, applications, approvals, status changes, ownership transfers, production history, completion reports, cost estimates — all of it, in bulk, for free.

Almost nobody uses it commercially, because the raw files are built for compliance reporting rather than for answering a question like “which operator should I call this month?” The datasets do not reference each other cleanly, the same asset appears under different identifiers, and the most valuable signal is usually an absence rather than a value.

This guide explains how to read it anyway: what each dataset carries, the order signal arrives in, and — most importantly — where the record stops being evidence.

The one rule that governs everything

The record is authoritative about what was filed. It is silent about what was agreed.

Every useful technique below follows from that sentence, and so does every way people get this wrong. A permit tells you an operator asked permission. An approval tells you permission was granted. A completion report tells you work finished. None of them tells you who was hired, what they were paid, or whether anyone has been hired at all.

Hold that line and the data is extremely useful. Cross it — by reading “no completion report” as “open for bid” — and you will build a pipeline out of clerical lag.

What the datasets actually carry

GOMDecom monitors 19 of them. They fall into five jobs.

Wells

Abandonment permits are where well decommissioning becomes visible first. An operator files an application to modify a well, including to plug it. This is normally the earliest hard signal that a specific well is heading for retirement, and it often appears well before any contract exists.

Borehole status carries the current state of every well — producing, temporarily abandoned, permanently abandoned. It is how you separate finished work from pending work.

End-of-operations reports confirm work finished. This is the dataset that closes a well out.

Watch the temporary-abandonment state carefully. A well sealed with recoverable barriers was never meant to stay that way, and some Gulf wells have been in that state for decades. Every one is future permanent-abandonment scope that a producing-wells count will miss entirely.

Structures

Platform structures lists what is standing. Platform removals carries removal applications and, when the work happens, the removal date.

The gap between those two is the signal. An approved removal application with no removal date on record is a structure that someone has committed to taking out and has not yet taken out. The full application-to-clearance lifecycle walks the filings in order.

One trap worth knowing: a removal application can also be closed out by a site-clearance date rather than a removal date. Read only one of those fields and you will count finished work as open.

Pipelines

Segment records and decommissioning permits sit in their own files, separate from wells and platforms, even when they belong to the same campaign. That separation is exactly why pipeline scope gets missed — we found over a thousand segments attached to non-completed Gulf campaigns.

Ownership and lifecycle

Lease assignments record approved transfers. Stacked up, they reconstruct a lease’s ownership history, which is what tells you who else remains exposed if the current holder fails.

Lease ownership and production give you the lease’s status and its last producing month. A terminated lease starts a decommissioning clock. A lease that stopped producing years ago is a candidate for idle-iron attention.

Cost

Four separate cost-estimate files, broken down by well, platform and lease. These are regulatory liability estimates expressed as percentiles — and the percentile you quote changes the number dramatically. They are not tender values.

The order signal arrives in

Decommissioning signal is sequential, and knowing the sequence is what lets you act early rather than read history.

1. Ownership moves, or production stops. Nothing has been filed. The lease has simply become a candidate. This is the earliest and weakest signal.
2. The lease terminates. A clock starts. Now there is a date attached to the obligation.
3. A permit or application is filed. The operator has committed on paper. This is the first hard signal, and commercially it is the most valuable moment, because the work is real and has not started.
4. Approval is granted. Timing firms up.
5. Work commences. A rig or vessel is on location. For most service lines the window has closed; for the ones that come later in the sequence, it is just opening.
6. Completion reports land. That scope is finished.

The commercially interesting band is 3 to 5 — filed, approved, not yet complete. Reading the signal chain covers the mechanics; the sales window covers how to time a call against it.

A caution on step 6: completion reports are filed by people, and people file late. The gap between work finishing and the record showing it is real and variable.

Reading absence

This is the hard part, and it is where most analysis of this data goes wrong.

Most of the value is in things that are not in the record. A well with a filed permit and no completion report. A structure with an approved application and no removal date. A lease past its deadline with scope still standing.

Each of those is genuinely interesting. But each has at least two explanations:

What you see	Could mean	Could also mean
Permit filed, no work commenced	Work has not started	It started and the filing is late
No completion report	Scope is open	Finished, report not yet processed
No rig identified	No contractor engaged	A contractor is engaged, and permits do not name them
Deadline passed	Operator is behind	A case-specific schedule was granted

You cannot resolve these from the data. So do not pretend to. The honest formulation is remaining observable scope — what the record does not show as finished — and it should never be relabelled as “unawarded”, “open” or “available”.

That is not pedantry. It is the difference between a prospect list a BD team trusts and one they stop opening.

Cross-checking

A single dataset will mislead you. Two datasets that agree are usually telling the truth.

• Permit against borehole status. A filed abandonment permit plus a borehole still showing as unplugged is a far stronger signal than either on its own.
• Removal application against structure list. An approved removal plus a structure still listed as standing is live scope.
• Lease status against production. A terminated lease that stopped producing years ago has both a clock and a reason.
• Cost estimate coverage. A campaign with cost estimates on file is better characterised than one without — not because the number matters, but because its presence means the regulator has assessed it.

Corroboration is also what should drive any confidence measure you build. Our own methodology scores evidence count and cross-dataset agreement separately from commercial attractiveness, precisely so a large opportunity backed by thin evidence does not outrank a smaller one that is well documented.

Identifiers, and why joining is hard

The datasets do not share one clean key. Wells carry API numbers, structures carry their own identifiers, pipelines carry segment numbers, and leases carry lease numbers that change hands over time. An area and block will group assets geographically but will not tell you they belong to the same commercial campaign.

Practical guidance:

• Join on the lease, then group by operator and area/block. That approximates a campaign better than any single identifier.
• Expect name variation. The same company appears under several spellings and several legal entities. Our operator pages show the parent behind each filing name for this reason.
• Keep the source values. Whatever you derive, keep the original record attached, so any figure can be traced back and checked.

We wrote up what building this actually involved, including the parts that did not work.

What the record can never tell you

Worth stating plainly, because it bounds every use of this data:

• Whether a contract has been awarded
• Who the contractor is
• What anyone was paid
• Whether a tender is open or planned
• What the operator intends to do next

If you need those, the public record is not your source. What it gives you instead is earlier: the scope, the timing, the deadline pressure and the counterparty — usually months or years before a tender exists.

Where to go next

• The vocabulary — the glossary defines every term here.
• The wider picture — the complete guide to Gulf decommissioning.
• Worked end to end — a full opportunity brief, with the source record behind every figure.

And the standing caution: verify anything you intend to act on against the government source. We are not the authoritative record, and neither is any product built on top of it.