CGM Accuracy Explained: What MARD Means for Your Readings
“Education is not a part of the treatment of diabetes”
MARD (Mean Absolute Relative Difference) measures how close a CGM's readings are to a lab blood test — lower is better, and modern sensors sit around 8–10%. A CGM is accurate enough to trust for trends and patterns, but it reads interstitial fluid, so it lags your actual blood glucose by a few minutes.
New CGM users often notice their sensor doesn't exactly match a finger-stick, and worry the device is broken. It usually isn't. Understanding one number, MARD, and one quirk, the lag, explains almost every discrepancy you'll see.
What MARD actually means
MARD stands for Mean Absolute Relative Difference. In plain terms, it's the average percentage gap between the sensor's readings and a reference lab measurement. A MARD of 9% means the sensor's readings are, on average, about 9% off from the true value. Lower is better. For context, today's over-the-counter sensors report figures in that range, for example around 8.3% for Dexcom Stelo and around 9.3% for Abbott Lingo. A decade ago these numbers were much higher, so modern consumer CGMs are genuinely accurate tools.
Why "9% off" is fine for what you're doing
A single reading being a few percent off sounds concerning until you remember what a CGM is for. You're not dosing insulin off it as a non-diabetic weight-loss user; you're watching trends, patterns, and how foods and the drug affect your curve. For that job, a MARD under about 10% is plenty accurate. The shape of your day and the direction of your glucose are reliable even if any single number carries a small margin.
The lag: why your sensor and finger-stick disagree
Here's the other big source of confusion. A CGM measures glucose in the fluid between your cells (interstitial fluid), not directly in your blood. Glucose reaches that fluid a few minutes after it reaches your blood, so the sensor runs slightly behind. When your glucose is changing fast, right after a meal or during exercise, a finger-stick and a sensor can disagree simply because they're reading two moments a few minutes apart. When your glucose is stable, they line up much more closely.
When to trust which
Trust the CGM for trends, patterns, and the overall picture. If you ever need a precise number for a real decision, for example if you feel low and need to confirm, a finger-stick blood check is the more exact reference in that instant. The two tools complement each other: the sensor for the continuous story, the finger-stick for a precise spot check.
Getting the most from an imperfect-but-useful tool
A CGM is a trend instrument, and reading it as one prevents most frustration. Don't over-react to a single reading or a brief disagreement with a finger-stick. Zoom out to the pattern across hours and days, which is where the accuracy and the value both live.
Read the trends, not the noise
Accuracy is about trends, and trends are exactly what's easy to misread reading raw numbers. Endobits turns your sensor stream into the reliable patterns MARD is good at, so you focus on what the device does well. Start with Endobits →
Sources
- Manufacturer accuracy (MARD) figures for Dexcom Stelo and Abbott Lingo, current as of mid-2026; verify at publish.
- Danne T, Nimri R, Battelino T, et al. International consensus on use of continuous glucose monitoring. Diabetes Care, 2017. 10.2337/dc17-1600