Garmin FTP auto-detect vs the guided test: what each number measures

Three numbers, three sources

Garmin's Edge manuals describe three routes to FTP: a profile estimate from your setup information, a guided FTP Test, and Auto Detect FTP, which learns from hard rides. Garmin says a more accurate value needs a paired power meter and heart rate monitor [Garmin Edge 1050 FTP], and both other routes list that pairing as a prerequisite.

The profile estimate is the number most new owners see first. The Edge uses the user profile information from the initial setup to estimate FTP [Garmin Edge 1050 FTP], which means it is built from who you are, not from anything you have ridden. Treat it as a placeholder. A rider at 70 kg who sustains 280 W and a rider at 70 kg who sustains 190 W can enter identical profiles.

The guided test is a structured effort the Edge leads you through. On the Edge 540 the path is My Stats, then Power, then FTP, then FTP Test, then Ride [Garmin Edge 540 FTP Test]. Once you start the timer, the device shows each step of the test, its target and your current power, and a message appears when the test is complete. You save the ride and the result appears in watts, in watts per kilogram and as a position on a color gauge.

Auto Detect FTP is the passive route. You select it from the more options menu under My Stats, Power, FTP, then ride at a steady, high intensity for at least 20 minutes outdoors and save the ride [Garmin Edge 1050 FTP]. Garmin adds a warning worth taking literally: the estimate may seem inaccurate at first, because the device requires a few rides to learn about your cycling performance. On Garmin watches, Auto Detection is on by default and covers maximum heart rate and lactate threshold as well as FTP when a compatible power meter is paired [Garmin Auto Detection].

When each number drifts

Auto-detect moves only when you hand it an effort near threshold, so easy riding leaves it frozen. A power meter swap can shift every number by several percent with no change in fitness. Even a perfect 20-minute test has error bars: against maximal lactate steady state, FTP's 95% limits of agreement run to about 9% [Borszcz et al. 2019].

No hard efforts is the most common cause of a stale number. Garmin's own instruction for auto-detect is a steady, high-intensity ride of at least 20 minutes outdoors [Garmin Edge 1050 FTP]. A rider in a base phase who rides two-hour endurance rides at 65% of FTP gives the detector nothing to read, so the number holds where it was while fitness moves. The same steps say outdoors, so a rider who does every hard session on an indoor trainer is outside the conditions Garmin describes.

Heart rate is a second input. Both the guided test and auto-detect list a paired heart rate monitor next to the power meter [Garmin Edge 540 FTP Test, Garmin Edge 1050 FTP]. Garmin does not publish how the two signals combine, so the practical rule is narrow: if the strap dropped out or read high during the effort that set your FTP, distrust that update rather than your legs.

A power meter swap is the drift nobody sees coming. When 54 power meters from 9 manufacturers were checked against a first-principles treadmill model, the mean deviation was -0.9% with a standard deviation of 3.2%, and 6 meters deviated by more than 5% [Maier et al. 2017]. Move from a crank meter to pedals or to a smart trainer and your FTP can step 10-15 W at 250 W with no change in your fitness. Any detector, Garmin's or ours, sees that step as real until you tell it otherwise.

The test itself has limits too. A 20-minute FTP agreed with maximal lactate steady state at the group level, with a bias of 1.4% and 95% limits of agreement of about 9.2%, and the agreement was wider in the less-trained group at 11.8% [Borszcz et al. 2019]. In recreational cyclists, a second study found FTP sat significantly below the lactate threshold, with a mean gap of 6.5% [Valenzuela et al. 2018]. A guided test number that reads a little low on a rider new to structure may be the protocol, not the rider.

How to turn auto-detect off, and when to

On Garmin watches the switch is under Physiological Metrics, then Auto Detection, where you choose the option you want [Garmin Auto Detection]. On an Edge, Auto Detect FTP sits in the more options menu under My Stats, Power, FTP [Garmin Edge 1050 FTP], so that is the menu to open when you want to stop it.

On a watch, hold MENU, select Physiological Metrics, then Auto Detection, then select an option [Garmin Auto Detection]. The manual states the feature is on by default, so a rider who never opened this menu has it running. The same menu holds maximum heart rate and lactate threshold detection, which matters for heart-rate zones as much as FTP does for power zones.

Turn it off for three cases. You just changed power meters and want time to compare the two before any number moves. You are racing a block of short, hard events where a single fast effort on a tailwind could raise FTP above what you can hold in training. Or you are coming back from illness, and you would rather set FTP by hand than let one hard day set it for you.

Leave it on when you ride hard efforts often, ride one power meter, and want a number that moves without a scheduled test. Run the guided test when you want a clean reading under controlled conditions: the same flat road or trainer, the same meter, rested. Whichever you pick, change one thing at a time. A new meter and a new test protocol in the same week leave no way to tell which one moved the number.

A third option: an estimate from the power curve

A power-curve estimate fits FTP to your best efforts across many durations rather than to one test or one ride. Critical power from a three-effort test and FTP from a 20-minute test correlated strongly, yet the 95% limits of agreement ran from -19 to 33 W, so the authors advise against using the two interchangeably [Karsten et al. 2021].

That gap is why different apps show you different FTP numbers for the same rider, and we cover the full comparison in our guide on why cycling apps show different FTP numbers. The short version: a guided test is one effort on one day, auto-detect is Garmin's read of recent hard rides, and a curve fit uses every best effort in the window. None of them is the true number. Each is an estimate with its own error bars.

AdaptCycling estimates FTP from the power curve of the rides you record, with no forced test. We fit a critical power model to your best efforts from the last six weeks and also check your best 20 minutes. A rise is proposed only after more than one ride backs it up. A drop is put to you for review rather than applied, and a jump of 15% or more is never proposed automatically, because that size of change is more often a power meter swap than fitness. With a Garmin, the estimate starts from the rides you record after you connect; past history arrives only through intervals.icu or a Garmin Connect export upload.

None of this replaces the wider question of how a Garmin fits into a self-coached season. Our guide to training with a Garmin covers which features need a power meter and where the plan layer stops. FTP is one input to that plan, and it is worth knowing which of your three numbers the plan is reading.

Quick answers

Sources cited in this guide

1. 01
   Garmin Edge 1050 FTP. Getting Your FTP Estimate (Edge 1050 Owner's Manual). Garmin.
2. 02
   Garmin Edge 540 FTP Test. Getting Your FTP Estimate (Edge 540 Owner's Manual). Garmin.
3. 03
   Garmin Auto Detection. Detecting Performance Measurements Automatically (fenix 6 Pro Series Owner's Manual). Garmin.
4. 04
   Borszcz et al. 2019. Is the functional threshold power interchangeable with the maximal lactate steady state in trained cyclists?. International Journal of Sports Physiology and Performance.
5. 05
   Valenzuela et al. 2018. Is the functional threshold power a valid surrogate of the lactate threshold?. International Journal of Sports Physiology and Performance.
6. 06
   Karsten et al. 2021. Relationship between the critical power test and a 20-min functional threshold power test in cycling. Frontiers in Physiology.
7. 07
   Maier et al. 2017. Accuracy of cycling power meters against a mathematical model of treadmill cycling. International Journal of Sports Medicine.

More inside Training with a Garmin

Other articles in this series

1. 01

   How to send a training plan to a Garmin Edge

   TrainingPeaks, TrainerRoad, JOIN, intervals.icu and Garmin's own builder: how far ahead each sends workouts and what happens when the plan changes.

   →
2. 02

   Getting old Garmin rides into a training app

   Why a new Garmin connection starts with no past rides, and the two routes that bring your history in: intervals.icu or a Garmin export.

   →
3. 03

   Garmin Unproductive status for cyclists: what to ride next

   Why Garmin says Unproductive while you get fitter on the bike, and four rules for what to ride next.

   →
4. 04

   Garmin Edge training plans: every option compared

   Garmin Cycling Coach, suggested workouts, TrainerRoad, TrainingPeaks or AdaptCycling on a Garmin Edge: what each needs, how it adapts, what it costs.

   →
5. 05

   Garmin workout not showing on your Edge or watch? The causes

   Why a planned workout misses your Garmin Edge or watch: device sync, the service's window, unsent workout types and the import permission.

   →
6. 06

   Garmin Cycling Coach without a power meter: what you get

   What a heart-rate-only cyclist gets from a Garmin, what needs a power meter, and how to train well on heart rate zones alone.

   →