A refrac is the cheapest well you'll ever drill — if you pick the right one. The problem is that the right one is buried in a book of hundreds or thousands of older wellbores, most of which have quietly settled onto a shallow tail and stopped asking for attention. The candidate you want is the well that produced poorly relative to what the rock can do, sits next to offsets that have proven the section, and has enough remaining life to justify the capital.
That's a screening problem, and it's exactly the kind of question you can ask the Wellsite data lake in plain language instead of building a spreadsheet from scratch.
The signature of a refrac candidate
Start with what separates a good candidate from a tired well that's simply done. Three things show up in the record:
- A low cumulative relative to vintage and area. A 2014 completion that cumed 180 MBO in a section where the county-average well of that vintage did 320 MBO is telling you something — either the original frac was undersized, the stage count was low, or the well was drilled before the operator dialed in the recipe.
- A steep, early flattening. Wells that fell off a cliff in the first 18 months and have been crawling along a thin tail often left oil behind rather than draining it. That's different from a well that declined gracefully along a normal b-factor curve and is genuinely depleted.
- Strong, more recent offsets. The single best evidence that rock will respond is a neighbor that already proved it — ideally a newer completion with a modern frac design outproducing the target by a wide margin.
None of those require a proprietary reservoir model. They're all in the production history.
Framing the ask
Instead of exporting decline data and grinding through it, you can pose the screen conversationally. Something like:
"In my book in Reeves County, find horizontal wells completed before 2016 whose cumulative oil is at least 30% below the county average for their completion year, that are still producing, and that have at least one offset within a mile completed after 2019 producing more than 2x the target's current rate."
That one question is doing a lot of work. It's pulling completion vintage, cumulative production summaries, decline state, offset geometry, and a benchmark against the county — and returning a ranked shortlist instead of a data dump. From there you narrow by hand.
Reading the shortlist
A shortlist isn't a work program. Each candidate still needs a look, and the record answers the follow-up questions fast.
Is it declining or already flat? Ask for the decline rate and the last twelve months of production. A well shedding 8% a year still has a curve to arrest; one that's been flat at 5 bopd for three years has less obvious upside and more mechanical risk.
How does it compare to its own offsets, not just the county? County averages smooth over a lot. Benchmarking the target directly against its nearest offsets tells you whether the underperformance is the well or the neighborhood. If every well in the immediate area is soft, the problem may be the rock, not the completion.
What did the operator learn since? If the same operator drilled newer wells nearby that stepped up hard, that's a proof point that their current design beats what went into the ground the first time. The Wellsite record lets you line up an operator's early completions against their recent ones in the same area and read the delta.
Why the tail matters
The most common mistake in refrac screening is chasing the lowest producers. A well making 2 bopd is cheap to ignore for a reason — often there's little pressure or reserve left to recover. The candidates that pay are usually the ones that are still making a respectable base, declined faster than they should have, and sit in delineated rock. You want underperformance against potential, not just low absolute numbers. Sorting the screen by the gap between the well and its vintage benchmark — rather than by raw rate — keeps you honest.
From screen to decision
Once you've got ten or fifteen names, the economics conversation gets simple. You know each well's remaining decline, its current rate, and the offset uplift you might reasonably capture. You can rank the candidates by the size of the recovery gap and by how strong the offset evidence is, then walk the list from best to worst until the capital runs out.
And because the same data lake tracks new permits and production changes, you can set an alert on the area — so if a competitor refracs a nearby well or a fresh completion lands next to one of your candidates, you hear about it before your AFE goes stale.
The takeaway
Refrac selection isn't about finding your worst wells — it's about finding wells that produced below what the rock has since proven it can give. That's a comparison the production record makes for you: vintage, cumulative, decline shape, and offset performance, ranked into a shortlist you can actually act on. Ask the question, read the tail, and let the neighbors tell you where the oil still is.