If you've ever looked at a continuous glucose monitor graph, the instinct is to focus on the highest point — the peak. A meal that pushes you to 150 looks worse than one that pushes you to 120. But the height of the spike is only half the story, and arguably not even the most important half.
Two curves, same peak, very different outcomes
Imagine two meals that both push your glucose up by the same 30 points above baseline. In the first, glucose returns to normal within 20 minutes. In the second, it stays elevated for 90 minutes before settling. Same peak height — wildly different total exposure.
This is the idea behind area under the curve (AUC), the standard metric researchers use to quantify glucose exposure in metabolic studies. AUC accounts for both how high glucose rises and how long it stays elevated. A tall, narrow spike and a short, wide plateau can have very different AUCs even with similar peaks — and AUC, not peak height alone, is what tracks most closely with total insulin exposure over time.
The key insight: insulin secretion responds to how much glucose stimulus your body has been exposed to over time — not just the single highest number on the graph. A curve that spikes and resolves quickly may produce a meaningfully smaller insulin response than a curve that rises less but lingers.
Why a spike alone can be misleading
There's a complication that makes peak height an even less reliable signal during exercise specifically: intense exercise can trigger your liver to release stored glucose into the bloodstream as part of a normal counter-regulatory response — completely independent of what you just ate. This means a hard workout right after a meal can sometimes push your peak higher, even while it's actively helping you clear glucose faster overall.
In other words, peak height during exercise reflects a mix of two different things: glucose from your food, and glucose your own liver added to the mix. Time spent elevated is a cleaner signal, because it's less affected by this confound — it more directly reflects how long your system actually took to bring things back down, regardless of where the glucose came from.
What this looks like in practice
Here's a real comparison from CostCarbs calibration data: across several no-exercise meals, glucose stayed elevated for an average of about 3.3 minutes per gram of carbohydrate eaten (above a modest threshold). One session that paired a mixed meal with a short, intense bodyweight workout brought that down to roughly 0.46 minutes per gram — an 86% reduction in total time spent elevated.
The peak itself in that exercise session wasn't dramatically lower than the no-exercise meals. What changed was how quickly the curve came back down. That's the part exercise is genuinely good at — and it's also the part that a peak-only view would have completely missed.
| Metric | What it tells you | Limitation |
|---|---|---|
| Peak height | How high glucose got | Can be distorted by exercise-induced liver glucose release |
| Time elevated | How long glucose stayed above a threshold | Less affected by short-term confounds; closer to total exposure |
| Area under curve (AUC) | Combines height and duration | Most accurate, but harder to calculate without continuous monitoring |
An important caveat
None of this means peak height is irrelevant — very high peaks matter for other reasons, including vascular stress and oxidative load. And the relationship between glucose AUC and insulin AUC, while well established in research as a general pattern, isn't a precise 1:1 translation. An 86% reduction in time-elevated does not necessarily mean an 86% reduction in insulin secreted — the true relationship depends on individual factors that vary person to person and likely isn't perfectly linear.
What the evidence does support is the general direction: less time spent elevated is consistently associated with lower total insulin exposure in metabolic research. That's a reasonable, evidence-informed goal to optimize toward — just not a number to treat as gospel from a single data point.
Why CostCarbs tracks time, not just peak
This is the reasoning behind shifting the CostCarbs calibration model toward time spent above a personal glucose threshold, rather than relying solely on how high the peak climbed. It's a more honest metric — one that's harder to distort with a single hard workout, and one that more directly reflects what you're actually trying to minimize: the total amount of time your body spends asking for insulin.