diff --git a/lightning_node_management.md b/lightning_node_management.md
index 343c424..14dc136 100644
--- a/lightning_node_management.md
+++ b/lightning_node_management.md
@@ -416,5 +416,36 @@ https://microlancer.io/task/view/2038
 The small node's pubkey:
 02c94a8c40c64c511228c0ca4b393ccd0738d75e9229073ceb6328246de32b9a8b
 
+## Getting scientifc with fees
 
+After going above 0.5 BTC in capacity, 20+ channel count and starting to use
+automated fee setting with charge-lnd, I have started to see an increase in the
+frequency and diversity of routed payments. It seems that the current size is
+getting to the point where routing activity flows in several directions so that
+channels stay balanced and things "flow nicely".
+
+Given that I don't expected to have any channel opening/closing activity for
+the next couple of weeks, I have decided I will run an experiment on prices for
+the next two weeks. The plan is to hold two different pricing models, each
+during one week, to see the impact on frequency and size of routing, as well as
+overall fees earned.
+
+The first week will be the "expensive" week. During that week, I will set:
+
+- The encourage-routing fee to 25.
+- The discourage-routing fee to 500.
+- The proportional fees between 100 and 300.
+
+The second week will be the "cheap" week. During that week, I will set:
+
+- The encourage-routing fee to 10.
+- The discourage-routing fee to 500.
+- The proportional fees between 15 and 50.
+
+Afterwards I will compare:
+
+- The count of forwards
+- The total forwarded amount
+- The total sats earned
+- The mean transaction size