Why I Chose One Formula and Refused to Add a Fallback

Most fitness apps let you skip the hard inputs. Don't know your body fat percentage? No problem — they'll estimate using your height and weight. You can be on your plan in under two minutes.

I deliberately didn't build that. The choice came down to two formulas — Katch-McArdle vs Mifflin-St Jeor — and the difference between them explains why MILA's outputs are more accurate for anyone outside the average body composition range.

Two Ways to Estimate How Many Calories You Need

BMR — Basal Metabolic Rate — is the number of calories your body burns at rest, before any activity is factored in. There are two main approaches to calculating it:

Mifflin-St Jeor is the most widely used formula in consumer health apps — considered the standard for general populations, though it falls short for anyone outside the average body composition range. Easy to implement. No additional inputs. It also assumes that body composition is predictable from height and weight alone.

It isn't.

MILA uses Katch-McArdle only. No fallback.

Why the "Easier" Formula Is the Less Honest One

The Katch-McArdle formula:

BMR = 370 + (21.6 × Lean Body Mass in kg)

Lean Body Mass = Total Weight (kg) × (1 − Body Fat %)

If you don't know your body fat percentage, you can't run this equation. That's not a bug. That's the point.

Consider two people:

Same weight. Same height, roughly. Nearly 300 calories apart in resting metabolism — before any activity multiplier is applied.

Mifflin-St Jeor would give both people nearly identical targets. Katch-McArdle would not.

⚠️ Weight-only formulas can miss true resting metabolic rate by more than 10% for people with high muscle mass or higher body fat percentages. For someone already frustrated by a plan that isn't working, that gap is often the difference between a plateau and progress.

When you use a formula that ignores body composition, you're not calculating for the person in front of you. You're calculating for a statistical average that may not resemble them at all.

The Fallback That Isn't a Fallback

I didn't remove the easy path. I replaced it with something better.

Instead of defaulting to a simpler formula when body fat % is unknown, MILA shows you three estimation methods at once:

All three show at once. You pick the method that fits what you know. If you've had a DEXA scan or use a body composition scale, you can enter that number directly.

Each method is labeled with its confidence tier. If the input is estimated, the output reflects that. No hidden assumptions.

💡 A reasonable estimate beats a precise guess at an incorrect value. Making you know your number — even approximately — produces a meaningfully better input than assuming you're average.

Named Levels, Not Black Boxes

Once TDEE (Total Daily Energy Expenditure — total calories burned per day at your activity level) is calculated, MILA offers four deficit levels for fat loss. Each has a name, a number, and an expected outcome:

You can see what you're choosing. You know what the name means, what the number is, and what rate of change to expect.

If you choose a pace where weekly loss approaches 1.5 lbs — the threshold where muscle loss risk increases — the app flags it. Not with shame. With information.

Every product decision in MILA has a reason. The formula, the estimation methods, the deficit levels — all of it is built to be visible, adjustable, and explainable. This is the core problem this whole series started with — the tools never show their work.

If you're the kind of person who wants to know the reason, read what I'm building next.