# Measuring a Setup Change

> Compare two confirmed runs of the same setup from your own telemetry — what the comparison needs, what it reports, and what it can never prove.

You changed the setup and drove it. Did it help? The objective comparison puts
two confirmed runs of the same setup side by side — a **baseline** and a
**test** — and reports what your own telemetry says moved between them.

> Setups is in **Public Beta** and requires a personal **Plus or Ultra** plan. A Team seat alone does not grant access.

## What You Need

- **Capture 1.3.8 or later.** The comparison needs the verified order of the
  captured laps. Telemetry from earlier versions of Capture cannot produce a
  trusted comparison.
- **Two confirmed runs of the same setup.** Each run is a confirmed association:
  a whole session or an inclusive lap range. The two runs can use different
  versions of the setup — for example, v1 as the baseline and v2 as the test.
  See [Setups and your sessions](/en/docs/ai-setups/sessions).
- **At least three eligible laps in each run.** With fewer, the result is
  **Inconclusive**.
- **At most 5,000 laps in each run.** Above that, Braking Lab stops before
  calculating from incomplete telemetry.
- **Available source telemetry.** If a session or a lap that defines a run was
  deleted, that run stays in your history but cannot be used for a new
  comparison.

## Opening a Comparison

Open the evaluation of a confirmed run — from **Setup used in this run** in the
session detail, or from a version's on-track evidence in the setup detail. Under
**Baseline vs test telemetry**, choose the **Baseline usage** and the **Test
usage**.

## What It Reports

| Field | What it means |
|-------|---------------|
| **Pace delta** | Test minus baseline. Negative means the test sample was quicker |
| **Consistency delta** | Test minus baseline. Negative means less lap-time spread |
| **Eligible laps** | How many laps each side contributed |
| **Status** | **Comparable**, **Inconclusive** or **Not comparable** |
| **Evidence** | **Weak** or **Moderate** |

Your own verdict sits beside the telemetry as **Driver thumb**, next to
**Objective telemetry**. The two are never merged: a faster run can still feel
worse, and a slower one can still be the setup you prefer.

**Recorded conditions** compare air temperature, track temperature, weather,
wetness, average fuel use, session type, fuel at the first selected lap, tire
compound, tire age and traffic. Each one reads **Similar**, **Different** or
**Unknown**. A condition missing on one side limits the evidence, and the
comparison names it.

**Evidence limitations** list everything that weakens the result: laps with
incidents that were excluded, laps whose sporting cleanliness was not measured,
run order or the time between runs, and more. Some limitations make the result
**Inconclusive**: fewer than three eligible laps, a run that may contain more
than one setup state, controls that changed without a usable timeline, or a lap
order that cannot be verified.

A pair is **Not comparable** when the simulator, car, car variant, track or track
configuration differ, when a usage is not confirmed, or when both selections
contain the same captured laps.

From the summary you can open representative laps of both runs in the telemetry
comparison.

## What It Can Never Prove

This is **personal A/B evidence** from your own runs. It is an association, not
proof that the setup caused the change. Track temperature, traffic, tire age and
your own driving all move between two runs. That is why Braking Lab never
declares a winner, and why the evidence stops at **Moderate** — two runs cannot
support a stronger claim.

Whether the setup is the reason stays your call. Your verdict is there for that.

## Saving a Comparison

Click **Save objective comparison**. The server recalculates both runs from your
current data and saves the result for your next iteration. Your feedback stays
separate.

If the telemetry of either run changes later, the saved comparison is marked as
stale. Click **Recalculate saved comparison** before you use it for an iteration.
If saving is temporarily unavailable, you can still review the comparison.

Your Race Engineer can compare two runs read-only at any time. To save a
comparison through the engineer, Braking Lab asks you to confirm directly
through your MCP client. This needs a client that supports MCP form elicitation.

## Measured Iterations

When you ask for the next change, your Race Engineer reads your saved
comparisons first. If a saved comparison measured the version it is about to
remix, it can link the new version to that measurement.

The new version then shows **Motivated by** with **Open source session**. That is
creation context only: it does not mean the new version was tested in that run.
To test the new version, drive it, confirm the run and compare again.

## Related

- [Setups and your sessions](/en/docs/ai-setups/sessions) — confirming and associating runs
- [Remixing setups](/en/docs/ai-setups/remix) — where the next version comes from
- [Comparing setups](/en/docs/ai-setups/comparing) — the parameter-by-parameter view of two setups