How We Test: Inside Our Instrumented Testing Protocol and Gear

Why we publish this

Every number in our instrumented testing section rests on a method. If you can't see the method, you can't judge the number — so this is the complete reference document for how we measure cars, what equipment we use, and what we record.

It exists for three reasons. Transparency: you should be able to interrogate our figures. Repeatability: another tester should be able to replicate our results. Accountability: publishing our protocol means we can be held to it.

This page is linked from every instrumented test we publish.

The core principle: conditions are part of the result

A performance figure without its conditions is close to meaningless. A 0–60 time recorded at sea level on a warm day with a prepared surface is a materially different measurement from one recorded at altitude, in the cold, on cold tyres.

So every test we publish carries a conditions block recording:

• Date and time
• Location and elevation
• Ambient temperature and, where relevant, surface temperature
• Barometric pressure and humidity
• Wind speed and direction
• Surface type and condition
• Tyre specification, age and pressures
• Fuel level or state of charge
• Occupants and payload

If a conditions block is incomplete, the test is not published.

Acceleration testing

| Measurement | Protocol | |---|---| | 0–60 mph / 0–100 km/h | Best of multiple runs, both directions, GPS-timed | | 0–100 mph, 0–130 mph | Where safe and appropriate to the car | | 5–60 mph rolling | From a rolling start, no launch control | | 50–75 mph / 40–80 km/h | The overtaking measurement — arguably the most useful figure we record | | Quarter mile | Elapsed time and trap speed | | Repeated runs | Minimum five consecutive, recording each |

Two protocol points deserve emphasis.

Trap speed matters more than elapsed time for assessing power. Quarter-mile ET can be flattered by traction and launch software; trap speed reflects how hard the car is still pulling after 400 metres. When verifying a power claim, we trust trap speed.

We publish the fifth run, not just the best. A single heroic launch measures peak capability in ideal conditions. Five consecutive runs reveal thermal management, and whether the car can actually sustain what it claims. This matters especially for EVs, where output can derate significantly with heat and state of charge.

Braking testing

| Measurement | Protocol | |---|---| | 70–0 mph | Maximum-effort stop, ABS engaged | | 100–0 mph / 100–0 km/h | Where safe | | Six consecutive stops | Recording each, with cooling intervals standardised | | Outputs | Best stop · average of six · degradation from first to sixth |

The six-stop sequence is non-negotiable. As set out in our braking test, the first stop measures the best case; the sixth measures what happens when you genuinely need it. Fade is where good cars and adequate cars separate, and it never appears in a single figure.

Roadholding and handling

• Skidpad: maximum sustained lateral acceleration, 300-ft circle, both directions, averaged
• Recorded alongside: tyre specification, because tyres frequently explain most of the result

EV-specific measurements

| Measurement | Protocol | |---|---| | Real-world range | Fixed mixed route plus a separate constant-speed highway run (75 mph / 120 km/h) | | Highway route | 200 miles, run in both directions to cancel prevailing wind | | Climate control | On, at 21°C — never switched off | | Charging curve | 10% to 100%, recording power every 1%, with and without preconditioning | | Key charging outputs | 10–80% time · average kW across 10–80% · miles added in 10 and 20 minutes | | Seasonal | Repeated summer and winter where possible |

The climate-control rule matters. Some published range tests switch off heating or cooling to maximise the figure. That produces a number no owner can replicate. We test at a temperature people actually sit in — 21°C — in every season and every market.

Everyday metrics

Cabin noise: dBA at a fixed position at driver's ear height, measured at idle, 50 mph and 70 mph, on a recorded surface type. Coarse and smooth surfaces differ by several dBA, so surface is always stated.

Boot space: measured in standardised suitcases, not litres. Carry-on cases (55 × 40 × 20 cm) and large cases (75 × 50 × 30 cm), counted seats-up and seats-folded, plus aperture dimensions, load-lip height and flat-fold assessment.

Fuel economy: brim-to-brim over a fixed mixed route, plus a separate constant-70-mph run. PHEVs are tested three ways — starting charged, starting depleted, and at a realistic weekly average — and we publish the depleted figure.

Weight: on our own scales, with front/rear distribution.

The equipment

[Complete with your specific equipment — this section should list actual hardware for full transparency]

• GPS-based performance timing system — model and accuracy specification
• Calibrated sound level meter — model, class rating, calibration date
• Corner-weight scales — model and capacity
• OBD/CAN data logger where the vehicle permits
• Weather station — temperature, pressure, humidity, wind
• Surface thermometer
• Tyre pressure and depth gauges
• Standardised luggage set for boot measurement

Calibration records are maintained and available on request.

What we don't do

We don't publish figures we haven't verified. Where a number comes from a manufacturer claim or another outlet's test, it is attributed in the results table, not presented as ours.

We don't cherry-pick conditions. If a car tested poorly in the conditions available, we publish that alongside the conditions rather than retesting until we get a flattering number.

We don't switch off climate control, remove weight, or over-inflate tyres to improve results.

We don't test on prepared drag surfaces and present the figures as representative. Where a surface is unusually grippy, we say so.

We don't hide the fifth run.

When we get it wrong

Instrumented testing involves equipment, conditions and human judgement, and errors are possible. If we publish a figure we later determine to be wrong, we correct it visibly — with a dated note explaining what changed and why — rather than quietly amending the page.

Equally, scores and figures can change legitimately. As set out in our re-testing policy, software updates can materially alter a car's behaviour, and we re-test when they do.

The bottom line

Numbers only mean something if you know how they were produced. This protocol exists so that every figure we publish can be interrogated, replicated and — where necessary — challenged.

The short version: we record conditions always, we test repeatedly rather than once, we publish the fifth run and the sixth stop, we keep the climate control on, we measure boots in suitcases, and we attribute anything that isn't ours.

If you spot something in our testing you think is wrong, tell us. That's rather the point of publishing this page.

• Every test publishes a full conditions block — a figure without conditions is close to meaningless
• We record five consecutive acceleration runs and six consecutive braking stops, because sustained capability matters more than a single peak
• Trap speed is trusted over quarter-mile ET for assessing power, as launch software can flatter elapsed time
• EV range is tested on a fixed route in both directions with climate control on at 21°C — never switched off to inflate the figure
• Figures sourced from manufacturers or other outlets are attributed in the results table, never presented as our own measurements

Key takeaways

• Every test publishes a full conditions block — a figure without conditions is close to meaningless
• We record five consecutive acceleration runs and six consecutive braking stops, because sustained capability matters more than a single peak
• Trap speed is trusted over quarter-mile ET for assessing power, as launch software can flatter elapsed time
• EV range is tested on a fixed route in both directions with climate control on at 21°C — never switched off to inflate the figure
• Figures sourced from manufacturers or other outlets are attributed in the results table, never presented as our own measurements