There's a lengthy technical paper which covers the subject:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2903977/
It's all way over my head, but there's two useful take home points.
Inserting a sensor and the body's response to it causes localised trauma followed by repair, and a foreign body response: both of those mechanisms chew up glucose and take time to settle, so dingbat results if you activate sensor too soon - it needs time to settle.
Also explains generally the "plumbing" layout: carbs into stomach, then to glucose in arteries, then distributed from arteries into smaller arterioles and capillaries, then seeps out from those around the body into the interstitial fluid surrounding cells, then taken into the cells to be used as energy. That process takes time, so meter testing blood directly from capillaries is unlikely to be the same from sensor testing the interstitial fluid - there is a relationship but they'll rarely be the same, especially when things are changing rapidly, most likely to be closest when things are settled.
A rough analogy might be comparing speed and volume of traffic on a motorway (blood) and that on the roads leading from the off ramps (interstitial). Related but not the same. The trick is figuring out how to account for the differences.
No-one ever said measuring glucose is an exact science!