Designing Coaching That Drivers Actually Follow

Most coaching advice fails on followability, not on accuracy. A driver who walks away from a session with “be smoother on entry” has not received coaching. They have received an observation. The gap between observation and prescription is the gap between a coaching surface that produces progression and one that produces good intentions. Closing that gap is the design problem the rest of this post is about.

This is the founder voice on the framework that the driver development plan article develops in technical detail. The doctrine claim, in one sentence: a followable prescription has anatomy, and any component missing from the anatomy is what makes the coaching go unfollowed.

Why “be smoother on entry” is not coaching

The advice “be smoother on entry” is observation-shaped. It names a quality the lap time would benefit from but not a procedure for producing the quality. The driver cannot act on it because nothing about it tells them what to do differently on the next lap. Similar examples: “carry more speed through the apex”, “trust the car on exit”, “be more committed on the brake”. Each one names a desirable end state. None of them name the action that gets the driver from where they are to that end state.

Driver-side, the failure mode is consistent. The driver nods, attempts to be smoother, has no procedural anchor for what smoothness consists of in this corner, and defaults to driving the corner the same way they did before. The coaching was technically correct. It was also not coaching, because it could not be followed.

The four-component prescription anatomy

A followable prescription has four components, each one load-bearing. Strip any one of them and the prescription collapses back into observation.

The weakness shape names the specific pattern observed in the trace — for instance, a two-stage trail-brake release with a mid-corner pressure spike, isolated to one corner of the lap. I use this example throughout the post because the trail-brake technique has a clear anatomy of its own — turn-in pressure, taper profile, release point — which gives a coaching prescription a precise vocabulary to land in. The drill name is the explicit action with a procedure attached: “hold trail-brake pressure on a single continuous taper from turn-in to 50% rotation, no secondary pressure application”. The success criterion is the measurable signal that says the drill worked: brake pressure shows a monotonic decrease from turn-in to apex, no local maximum past 50% of the trail-brake duration. The review cadence is when the driver checks back: ten laps, then re-diagnose.

A prescription missing any one of these is observation- shaped, not prescription-shaped. The drill name without a success criterion produces infinite drilling without a finish line. The success criterion without a review cadence produces a drill that runs until the driver gets bored. The weakness shape without a drill name produces frustration without an action — the driver knows something is wrong but has nothing to do about it.

The five verbs run as a loop

The five framework verbs — diagnose, prescribe, execute, measure, adapt — are not a line. They are a loop. Order matters: diagnosis precedes prescription, prescription precedes execution, and the loop only closes when the adaptation step feeds back into a re-diagnosis.

This is what the doctrine names “the loop closes or the feature does not ship”. Coaching that emits a single prescription and walks away produces good intentions. The driver runs the drill, has no signal for whether it worked, and either keeps running the same drill indefinitely or abandons it without knowing whether it produced the result. Coaching that runs the loop produces follow-through: the driver runs the drill, measures against the success criterion, adapts on the next session, and the diagnosis layer either retires the prescription or surfaces a new one.

Why a hand-curated rule library makes prescriptions followable

Every coaching prescription in LAP traces to a written rule that a human author published their name against. The reason a hand-curated rule library matters for followability is not credentialism. It is that the rule library is the source from which the four prescription components are reliably populated.

A rule in the library names the weakness shape it diagnoses, the drill that fixes the shape, the success criterion that says the drill worked, and the cadence at which to review. When a session’s diagnosis layer matches a rule, the prescription is complete by construction. The driver receives the four components together, in one piece, because the rule was written to ship them together.

When a session’s diagnosis matches no rule, the surface declines to prescribe rather than producing a fabricated prescription with one or more components missing. The deliberate scope choice is durability over surface area. A coaching surface that fabricates to fill gaps in the rule library produces non-followable advice and erodes trust the first time it lands on a corner the driver knows well. A surface that declines to prescribe when the library has not earned the right to produces fewer prescriptions, but every prescription it does ship survives contact with the corner.

What this looks like in a session

A driver opens LAP after a session. The diagnosis layer surfaces three weakness shapes from the session’s data: a trail-brake release pattern at one corner, a throttle- pickup hesitation at another, and an apex-line inconsistency through a third. Each one comes with a drill name, a success criterion, and a review cadence. The driver picks one, runs the drill across the next ten laps of practice, checks the success criterion, and the adaptation step either retires the drill or feeds back into a re-diagnosis.

If the third weakness shape matches no rule in the library, the surface presents it as “observed pattern; no current prescription” rather than fabricating a drill name that will not survive contact with the corner. The driver knows what to work on this week without being handed coaching that cannot be followed because it was never followable to begin with.

The cross-platform discipline

The same prescription anatomy runs on sim and on real-world surfaces. The architecture that makes one coaching surface ship across both — the cross-platform Driver OS — is the same one that the sim-to-real transfer article develops in detail, and the magnitudes-do-not-transfer caveat that the founder-voice version develops in plainer language. A weakness shape extracted on iRacing on Tuesday produces a drill name, a success criterion, and a review cadence. The driver runs the drill on Saturday at the track and adapts. Same anatomy, same loop, magnitudes recalibrated on the day.

Without that unification, a driver who runs both surfaces receives two disconnected coaching streams — one for the sim, one for the real car — and has to reconstruct the connection in their head every session. With it, the prescription anatomy is the connection.

Why followability is the design constraint

Coaching that is accurate but not followable produces no progression. Coaching that is followable but inaccurate produces drills that fix nothing. The design target is both — accurate diagnosis matched to a followable prescription, with the surface declining to prescribe when the rule library has not earned the right.

Followability is not a marketing line in the Driver OS. It is the design constraint that determined how the rule library is structured, why the prescription anatomy has four components rather than one, and why the loop must close before a feature ships. Coaching is followed when it tells the driver the next four things in the right order and gives them a way to know whether the drill worked. The work between the diagnosis and the next session is where the lap time gets earned, and the coaching surface either equips the driver to do that work or it gets in the way.