October 17, 2026 · 10 min read

Is my endurance improving? Efficiency factor, cardiac drift and cadence

Fitness (CTL) tells me how much I train. The power curve tells me how much I can put out. But is my endurance improving — does the same work cost me less? Four cards on the Progress page answer that, spread across both tabs: efficiency factor, cardiac drift, heart rate trends and cadence efficiency. This is the fourth article in the series, after zones and intensity distribution.

Four cards, one question

What they have in common: all four need power and heart rate at the same time. Without a chest strap (or another heart rate sensor) none of them will fill in.

Efficiency factor: how many watts per heartbeat

EF is a ride's normalised power divided by its average heart rate [1, 2]. If I ride 130 W at 140 beats per minute, EF is 0.93. If two months later the same power costs me 130 beats, EF rises to 1.0 — the heart does less work for the same watts. The app computes EF for every ride that lasts at least 20 minutes and has both power and heart rate, and draws a trend line through the points.

My trend line rises by 0.0072 a week, which is 0.9% of my average EF. The app calls the trend “Improving” when the change exceeds 0.5% a week, and “Declining” when it is as large the other way. The median in April was 0.77, in June 0.89, in October 0.93. This is exactly the effect an aerobic base works towards.

The two highest points are ramp tests — and it's worth understanding why. In a short, ever-harder effort, power rises faster than heart rate, which doesn't catch up with the watts before the test ends. A high EF in a test does not mean I was in my best shape then. That is why the card says: compare rides of a similar type. Easy Z2 rides of similar length give the most honest picture. I write more about EF in the article on efficiency factor.

Below the chart the app says how many rides from the period did not make it onto the chart — 11 for me. Those are rides shorter than 20 minutes or without power or heart rate.

Cardiac drift: does your heart rate run away in the second half?

Cardiac drift (aerobic decoupling) splits a ride into two halves and compares the ratio of power to heart rate in each. If in the second half your heart rate is higher at the same power, the drift is positive: the heart has started working harder and endurance is running out [2]. For the result to mean anything, the ride has to be steady and long, so the app only computes it for rides that:

• last at least 30 minutes,
• have a variability index (VI) of at most 1.15 — so they are not intervals,
• have at least 80% of their time covered by power and heart rate data.

My average is 2.9%, and the trend is stable. The app only announces a change when the average moves by at least 2 percentage points, and it needs at least 6 rides for that. Heat, dehydration and fatigue raise drift regardless of fitness, so a single high result says little.

13 anomalies is more than I expected. Negative drift means that in the second half there was more power per heartbeat than in the first. For me these are almost exclusively recovery rides and very easy ones — “Active Recovery”, “Recovery Spin”, “Endurance ride”. At low power, heart rate is often still raised at the start (after the commute, a coffee, a day at work) and only settles later, and at 90–100 W a few beats of difference amount to several per cent. Such a result says nothing about endurance, which is why the app flags it separately. Z2 rides of an hour or more say the most about your base. I write more about the phenomenon itself in the article on cardiac drift.

Heart rate trends: high for you, not high in general

The Heart rate trends card shows the average and maximum heart rate of every ride against your moving normal — the average of your last 10 readings. The normal only appears after 4 rides; before that, a deviation from two readings would be noise with a number attached. There are no “too high” thresholds here: 150 beats can be an easy ride for one person and a very hard one for another.

For me the clearest signal is at the start of the season. In April and May a run of rides had an average heart rate 12–20 beats above my normal from the preceding days — exactly when TSB on the Form tab was falling to −37. The last ride before I wrote this, on 15 October, averaged 127 beats against a normal of 134. The card itself warns that it mixes different kinds of rides, so it's best read in runs: several rides in a row clearly above normal at a similar effort point to fatigue, illness or heat. When to back off training based on heart rate, I discuss in the article on HRV and resting heart rate.

Cadence efficiency: a description, not a target

The last card collects every second of riding in the period in which cadence, power and heart rate were all recorded, and splits them into 5 rpm bands from 50 to 120. For each band it computes watts per heartbeat. A band needs at least 3 minutes, and coasting seconds (below 50 W, heart rate below 80 or cadence below 30) don't count at all.

At first glance the result suggests I should ride at 55 rpm: there I get 1.06 W per beat, at my usual 85 rpm only 0.81. That would be the wrong conclusion. For me, low cadence mostly means climbs and hard stretches, where average power was higher (147 W at 55–60 rpm against 113 W at 85–90 rpm) and heart rate hadn't had time to rise with it. Cadence, gearing, terrain and intensity go together — the card shows where my riding came out best, not where it would come out if I changed one of those things. Research also shows that the most economical cadence rises with workload [3].

The most interesting thing about this card is something else: the distribution of time. 80 of my 112 hours with paired data are at 80–95 rpm — the cadence a trainer in ERG mode simply imposes. If you want to work on your cadence deliberately, start with the article on optimal cadence.

What I take away from this

In the last article of this series: power by grade and when I ride — terrain and habits.

Sources

1. Allen H, Coggan AR, McGregor S. Training and Racing with a Power Meter. 3rd ed. Boulder, CO: VeloPress; 2019.
2. Friel J. The Cyclist's Training Bible. 5th ed. Boulder, CO: VeloPress; 2018.
3. Foss Ø, Hallén J. The most economical cadence increases with increasing workload. Eur J Appl Physiol. 2004;92(4–5):443–451. doi:10.1007/s00421-004-1175-5

← All blog posts