diff --git a/Memory-footprint:-what-happens-inside-µBlock-after-installation.md b/Memory-footprint:-what-happens-inside-µBlock-after-installation.md
index f0b3a3c..4689f1e 100644
--- a/Memory-footprint:-what-happens-inside-µBlock-after-installation.md
+++ b/Memory-footprint:-what-happens-inside-µBlock-after-installation.md
@@ -20,10 +20,11 @@
 
 Now if you have reached the point where there is a valid µBlock selfie, this is when µBlock will run the most efficiently.
 
-The next time you launch µBlock and there is a valid selfie, the load time will be a fraction of the load time when no selfie is available, and µBlock's baseline memory footprint will be smaller than when µBlock launches without a selfie available.
+The next time you launch µBlock and there is a valid selfie, the load time will be a fraction of the load time when no selfie is available <sup>[1]</sup>, and µBlock's baseline memory footprint will be smaller than when µBlock launches without a selfie available.
 
 So my point is that µBlock will perform best efficiency wise if you leave it time to optimize itself after installation: In subsequent launches, µBlock will perform more efficiently than what you may have observed right after you installed it.
 
+<sub>[1] See <https://www.youtube.com/watch?v=BpypOeK10N8></sub>
 ***
 
 Also, if you want to look at memory figures, aside the above notes regarding garbage collection, keep in mind: