The Strength Work Your Running App Bolted On

Open your plan for next week. Somewhere between the Tuesday intervals and the Thursday tempo there is probably a little dumbbell icon labelled “Strength, 20 min” or “Mobility + Core.” Tap it. If what you find is three rounds of bodyweight squats, glute bridges, a plank and some mountain climbers, you have found the most common failure in running apps with strength training: the session exists so the app can tick a feature box, not so you can run better.

This post is about telling the difference. You already read your own Strava and watch data. You can apply the same scepticism to the gym sessions your app hands you.

The three tests

I judge any app-supplied strength session on three questions. Fail one and it’s decoration. Fail two and it’s a liability.

1. Is it specific? Does the session target what breaks runners (calves, soleus, glute medius, hamstrings, hip flexors, foot intrinsics) with exercises loaded enough to matter? Or is it a generic fitness circuit that could be handed to anyone?

2. Is it progressive? Does week 6 look different from week 1? Runners adapt to resistance work the same way they adapt to mileage: by being asked for slightly more. Strength is not a maintenance chore. A 20-minute session that never changes is a warm-up with extra steps.

3. Is it sequenced around hard runs? Where does it sit relative to your intervals, tempo and long run? Placement decides whether strength helps or quietly wrecks your key sessions.

Test one: specificity

A bodyweight single-leg calf raise is easy for most adults. The soleus and Achilles take real load when you run: roughly 6 to 8 times bodyweight through the Achilles at faster paces, according to the biomechanics literature that gets quoted often. Doing 15 unweighted raises does not tell that tissue anything new.

Compare two sessions. The first is what I pulled from a typical “runner strength” template:

Strength A (20 min)
Bodyweight squat      3 x 15
Glute bridge          3 x 15
Plank                 3 x 30 sec
Mountain climber      3 x 20
Calf raise            3 x 15

The second is what a coach who lifts would write for the same 20 minutes:

Strength A (25 min)
Goblet squat, 16 kg           3 x 8   (3 sec down)
Single-leg RDL, 2 x 12 kg     3 x 8 each
Step-up, 30 cm box, 8 kg      3 x 8 each
Single-leg calf raise, 
  straight knee, 10 kg        3 x 10 each (2 sec up, 3 sec down)
Seated soleus raise, 20 kg    3 x 12
Side plank with hip abduction 2 x 30 sec each

The first has no load, no tempo and rep counts so high that nothing is being strengthened; it’s mild cardio. The second uses heavy-ish sets of 8 to 10, slow lowering and single-leg work, which is where running actually lives. Notice too that mountain climbers are conditioning, not strength. They add fatigue without adding resilience.

Quick check: if every exercise could be done in a hotel room with no equipment and the app never asks what you own, it’s generic.

Test two: progression

Open week 1 and week 8 of your plan and compare. Here’s what honest progression looks like for the second session above across a 12-week block:

Load rises, reps fall, then volume drops with the taper. Nothing in that table needs an algorithm. A plain linear rule does it: add about 2 to 4 kg when you hit every rep with two reps in reserve.

What do the apps do? Runna’s strength sessions are built as separate guided workouts, and in my experience the exercise lists change less between weeks than the running does, with progression mostly coming through added sets or variations rather than prescribed weights. Garmin Coach, the free adaptive plan on Garmin Connect, is a running-only engine: it adjusts paces and distances from your watch data and has no strength progression to speak of, so any gym work you do sits outside it. Check the current state of both yourself, since these products change often, but the test stays the same: can you see the load rising?

ChatGPT-built plans are the wildcard. Ask for “a 12-week marathon plan with strength” and you will usually get a table where Wednesday says “Strength training (30 min)” in all twelve rows. That’s a placeholder, not a prescription. You can fix this by asking for specifics, which I cover below.

Test three: sequencing

This is where generic plans do real damage. Look at a typical week from an app:

Mon  Rest
Tue  Intervals: 6 x 800 m @ 5K pace
Wed  Easy 8 km + Strength
Thu  Tempo: 5 km @ threshold
Fri  Rest
Sat  Easy 6 km
Sun  Long run 18 km

Strength on Wednesday looks harmless. But heavy single-leg work on Wednesday gives you soreness peaking Thursday, when you’re asked to hold threshold pace. And the Tuesday intervals mean your legs were already cooked. Compare with a better week:

Mon  Strength A (heavy lower) 
Tue  Intervals: 6 x 800 m @ 5K pace
Wed  Easy 8 km
Thu  Tempo: 5 km @ threshold
Fri  Strength B (lighter, upper + core + calves)
Sat  Easy 6 km
Sun  Long run 18 km

Heavy work sits the day before intervals? Not ideal either, honestly. The cleaner version puts the heavy session straight after a hard run, same day if you can manage it (intervals in the morning, lifting in the evening, 6+ hours apart), so that the following day is genuinely easy. Hard days stay hard, easy days stay easy. The rule of thumb:

• Heavy lower-body strength: on or just after a hard run day, never the day before one.
• Never within 48 hours before a long run or race-pace session.
• Taper: drop to one light session in the final 10 days, nothing new, nothing sore.

Apps that drop strength into “rest days” are the ones to be wary of. A rest day is when the adaptation from your hard running happens. Adding a loaded session there removes the one day your legs had to recover.

How to audit your own plan in ten minutes

Do this with last month of your Strava and your plan open side by side.

1. Count the load. Does any exercise have a weight, or a rep range tied to a weight? If none, it’s a circuit.
2. Compare week 1 to week 5. If the exercise list is identical and the numbers unchanged, there’s no progression.
3. Map the soreness. In Strava, note your perceived effort on the run after each strength day. If your easy runs after strength sit 10 to 15 seconds per km slower at the same heart rate, the placement is costing you.
4. Check the heart rate drift. On the Thursday tempo after a Wednesday lift, did your heart rate run 4 to 6 bpm higher than your usual for that pace? That’s fatigue from the wrong sequencing.

Fixing a ChatGPT plan, with a prompt

If your plan came from a chatbot, don’t accept the placeholder. Paste something like this:

Rewrite the strength component of my plan. I have a 16 kg kettlebell,
a pair of 12 kg dumbbells and a 30 cm step. I run Tue (intervals),
Thu (tempo) and Sun (long). Place two sessions per week so that heavy
lower-body work is never the day before a hard run. Give sets, reps,
tempo and starting loads, and tell me exactly when to add weight.
Reduce volume by 50% in the final 10 days before my race on 14 April.

Then check the output against the three tests. A model will happily write a progression, but it won’t know that your left calf is cranky, so you remain the safety check.

What to do this week

Pick one app session and upgrade it. Take the calf raise, put a loaded backpack or a dumbbell in your hand, slow the lowering to three seconds and do 3 sets of 8 per leg. Move the session to the evening of a hard run day. Write the weight down. Next week, if it felt like two reps left in the tank, add 2 kg.

If you’re still choosing which app to trust, the head-to-head comparison in AI Coaching Apps, Tested Against Each Other covers how the running plans themselves hold up. Strength is just the part most of them leave thinnest.