Introduction
When a drill bit advances thousands of metres below the surface, nobody can watch it directly. The rock it passes through is invisible, the formation boundaries shift unexpectedly, and a single mis-turn can leave an expensive well outside its productive zone entirely.
Geosteering is the discipline that solves this problem. It is the real-time process of using subsurface data — captured while drilling is in progress — to adjust a wellbore's trajectory so that it stays within a target formation. Done well, geosteering increases hydrocarbon contact, improves completion efficiency, and directly drives well economics. Done poorly, it wastes millions of dollars in remediation or leaves pay untouched.
This guide covers what geosteering actually is, the tools and data that underpin it, and how modern AI platforms are changing the discipline.
What Problem Does Geosteering Solve?
Conventional vertical wells punch straight down through a reservoir. That is fine for thick, tabular formations with predictable geometry. But most reservoirs worth drilling today are thin, laterally variable, or structurally complex. Horizontal and deviated wells are the norm because they expose far more reservoir surface area to the wellbore — but only if the wellbore actually stays in the reservoir.
Reservoirs are not flat. Faults, folds, unconformities, and depositional pinch-outs can cause the target zone to rise, drop, or disappear entirely within a few hundred metres of lateral drilling. A pre-drill geological model is an interpretation based on sparse data. It is always wrong in detail, and frequently wrong in ways that matter.
Geosteering is the feedback loop that closes the gap between the pre-drill model and geological reality. It takes real-time sensor data from within the wellbore and uses it to update the geological interpretation continuously — enabling the directional driller and geologist to steer into, or back into, the pay zone.
Core Data Sources
Measurement While Drilling (MWD)
MWD tools sit behind the drill bit and transmit data to surface in real time via mud pulse telemetry, electromagnetic pulses, or wired drill pipe. The primary outputs are:
- Inclination and azimuth — the wellbore's 3D orientation at each survey point
- Tool face — the angular position of the motor bend, which controls the direction of the next deflection
MWD data defines where the wellbore actually is in 3D space. Without it, you cannot steer at all.
Logging While Drilling (LWD)
LWD tools measure formation properties in real time. The most important for geosteering are:
- Gamma ray (GR) — natural radioactivity of the rock. Shales read high; clean sands and carbonates read low. GR is the primary formation correlation tool.
- Resistivity — measures how strongly the rock resists electrical current. Hydrocarbons are highly resistive; brine-saturated rock is conductive. Resistivity is the primary fluid discrimination tool.
- Neutron and density — measure porosity and lithology. Useful for formation identification and fluid typing.
- Image logs (while drilling) — high-resolution electrical or acoustic images of the borehole wall. Used to identify fractures, bedding dip, and structural features.
Modern LWD tools can also measure azimuthal resistivity — the formation resistivity seen in different directions around the tool. This is the basis of look-ahead and look-around capability, which can detect bed boundaries before the drill bit reaches them.
Seismic Data
Pre-drill 3D seismic provides the structural framework for geosteering. The seismically interpreted surfaces define the expected position and dip of formation tops. Real-time LWD data then updates these surfaces as drilling reveals the actual geology.
Mud Logging
Mud logging captures cuttings (rock chips returned to surface in the drilling fluid) and hydrocarbon gas shows. It provides a continuous lithological record that is cheap and often overlooked, but valuable for ground-truthing the LWD interpretation — particularly when telemetry is lost.
The Geosteering Workflow
1. Pre-Drill Planning
The geologist and geophysicist build a geological model from seismic, offset well data, and formation evaluation reports. This defines the expected position of the target formation in 3D space. The directional driller designs a wellbore trajectory — typically an anti-collision-safe path with landing point, turn rates, and lateral azimuth.
2. Landing
As the well approaches total depth of the vertical section, the driller begins curving the wellbore toward the target formation. Gamma ray and resistivity responses are compared against the pre-drill model and offset well logs. The geologist picks the formation top in real time and calls the landing depth. The directional driller adjusts tool face to bring the well onto target.
3. Lateral Steering
Once the well is landed, the real challenge begins. Every few metres, the geologist reviews the incoming LWD data:
- Is the gamma ray increasing (approaching a shale) or decreasing (approaching a cleaner, better reservoir)?
- Is resistivity holding up (hydrocarbons) or dropping (water contact)?
- What does the depth-adjusted correlation with offset wells suggest about the formation dip?
Based on this interpretation, they call a steering decision: hold angle, build angle (steer up), drop angle (steer down), walk left, or walk right. The directional driller adjusts the tool face and executes the deflection on the next connection.
This cycle repeats every 30–90 minutes throughout the lateral section, which can be several kilometres long.
4. Real-Time Model Updating
As the well penetrates the formation, the true dip and geometry become clearer. A skilled geoscientist continuously revises the interpretation — adjusting the projected formation top, correlating with the nearest offset well, and deciding whether an anomaly represents local structure, a fault, or simply noise in the log.
This is where experience, pattern recognition, and the quality of the geological model converge. It is also where most geosteering errors occur.
What Makes Geosteering Difficult?
Telemetry Lag
MWD/LWD tools transmit data through the mud column. Mud pulse telemetry — the most common method — is slow: typically 1–12 bits per second. At fast drilling rates, the drill bit can be 50–100 metres ahead of the last data point. Decisions are made on information that is already partly outdated.
Formation Complexity
Real reservoirs are geologically messy. Thin interbeds, cemented zones, diagenetic alteration, and lateral facies changes can all produce log responses that look like formation boundaries but are not. Distinguishing a true geological boundary from a local heterogeneity requires judgement, not just pattern matching.
Depth Uncertainty
Wellbore positioning in 3D space accumulates error with distance from the last survey point. In long laterals drilled at low inclination, the vertical depth uncertainty can be several metres — enough to be ambiguous about which side of a formation boundary the bit is on.
Decision Pressure
Geosteering happens while the rig is running, which means every hour of lateral drilling costs tens of thousands of dollars. Decisions cannot wait for extended analysis. The geoscientist must interpret data quickly and communicate clearly to the directional driller under time pressure.
Modern AI-Powered Geosteering
From Spreadsheets to Platforms
For most of the industry's history, geosteering was done with spreadsheets, offset well log overlays printed on paper, and hand-drawn cross-sections updated with a ruler. The data was real-time; the interpretation workflow was not.
Modern geosteering platforms have changed this. They ingest WITSML data streams directly, display LWD curves alongside the trajectory in real time, allow digital formation top picking, and propagate the geological model update automatically. The geoscientist can see the well position, the updated geological interpretation, and the next steering call on a single screen.
Where GeoEngine AI Fits
GeoMaster's GeoEngine takes this further. It applies machine learning to the incoming LWD data to:
- Detect formation proximity before the bit arrives, using azimuthal resistivity gradients
- Correlate in real time against the stored offset well library, suggesting the most geologically similar reference well and aligning the depth scale automatically
- Flag anomalies — log responses that deviate significantly from the expected model — so the geoscientist's attention is drawn to decision points rather than routine sections
- Generate a steering recommendation at each connection, based on the current geological interpretation and the configured target formation
These are decision-support tools, not autopilots. The geoscientist remains in command. GeoEngine reduces the cognitive load of routine data processing so that human expertise is focused on the decisions that actually require it.
Key Performance Indicators
How is geosteering quality measured? The industry typically tracks:
| Metric | Definition |
|---|---|
| Target reservoir contact (TRC) | Percentage of lateral length drilled within the designated target formation |
| Net-to-gross (N/G) | Ratio of productive reservoir to total lateral length |
| Steering events per metre | Number of directional corrections per unit of lateral — a proxy for how much the formation deviated from the pre-drill model |
| Gamma ray average | Average GR in the lateral section — lower is generally better for clean sandstone reservoirs |
A well with 90%+ TRC in a thin reservoir is a geosteering success. A well with 60% TRC could have been drilled differently — and the difference is often the quality of the real-time interpretation.
Conclusion
Geosteering is one of the few disciplines in subsurface engineering that is simultaneously geological interpretation, real-time data analysis, and operational decision-making under pressure. It bridges the gap between the pre-drill model and geological reality, and it directly determines how much of the reservoir a well actually produces.
As reservoirs become thinner, formations more complex, and well economics tighter, the stakes for getting geosteering right are higher than ever. Platforms that bring real-time WITSML data, geological interpretation, and AI-powered decision support into a single workflow are not a luxury — they are the standard that modern drilling teams should be working toward.
To see how GeoMaster handles this in practice, start a free trial and connect your first WITSML feed.