Diagnostics over the parts cannon: how a fault should be found

P0171 — system too lean, bank one. Somewhere right now, a car with that code is getting a new oxygen sensor it does not need. The code does not say the sensor is bad. It says the sensor is reporting a condition the control module cannot correct. Those are very different sentences.

Proper diagnosis starts by reproducing the fault and watching live data while it happens: fuel trims at idle and at load, mass-airflow readings against calculated expectation, fuel pressure under demand. A lean condition that disappears at higher RPM points at an unmetered air leak. One that gets worse under load points at fuel delivery. The code is identical in both cases; the repair is not.

The second discipline is testing the circuit, not the component. A sensor that reads implausibly can be failing, or it can be a corroded connector, a chafed wire against a bracket, or a poor ground shared with three other modules. Backprobing the connector and checking voltage drop takes fifteen minutes and prevents a replacement part from inheriting the same fault.

Third: verify the repair the same way you found the fault. Clear the codes, run the drive cycle, confirm the monitors complete and the trims settle. A repair that has not been verified under the original conditions is a hypothesis, not a fix.

This is slower than swapping parts, and it is cheaper. Diagnostic time is the one line on an invoice that consistently pays for itself — usually several times over, in parts nobody had to buy.