I used the second half of this game to experiment with a Missile Silo, which means I have another batch of Lessons Learned.
1) I built the silo to solve a specific problem - Cerberuses under area shields with Racketeers preventing a direct assault. I partly accomplished the goal, destroying some Cerberuses but never all of them at once. The main impact of this was letting my team claim about 10% of the map's mexes, and hold a porc strongpoint, that would otherwise have been within Cerberus and radar range.
2) I used the combo straight off the wiki to kill Cerbs: 1 Eos to drain shields, then 2 more after a delay to get the kill. This worked but it's at a metal disadvantage if the other team reclaims the Cerb wreck and the Eoses don't kill something else with splash damage.
3) To maximize the chances of killing a secondary target with splash damage, I can place the Eoses off center on the main target. This is particularly important when the main target is on a short pedestal.
4) I should've been using more Shockley->Eos strikes against dense targets. The entire attrition advantage from my Silo use came from a single Shockley->Eos strike on a standing shield ball. Everything else combined was not worth the cost of ammunition from an attrition perspective.
5) Shockley has a larger radius than Eos, which implies that a single Shockley can be followed up with two Eos strikes on different stunned targets for better efficiency. Here, I had the second Eos and a surplus of predictably still-stunned targets but simply didn't think to take the shot.
6) An anti-shield Shockley creates two vulnerabilities, one immediate when it knocks out the local shield generators before Aspises have a chance to walk into the gap, and a second when the stunned Aegises come back online and drain the rest of the network. I should have been trying to exploit that second window also.
7) I was trying to support my jumpbot ally's advances with Ducks, Bouys, Bulkheads and Grizzlies, and generally losing them without accomplishing much. While this was partly due to lack of attention and auto-retreat could have helped, I should have considered spending some but not all of that support budget on anti-army Shockley strikes instead of units.
8) Swifts and Sparrows can scout past Anglers and other missile AA. The missiles eventually catch up but they've already seen what's behind the AA.
9) The opposing silo was getting much better return on Eoses than mine was, partly due to only forcing through shields for very high value targets like Big Bertha, but partly by hitting targets like a Merlin that I wasn't even looking for.
10) While shield linking creates the 3-Eos solution to a Cerberus, it also makes the window of vulnerability very narrow when there are 20+ linked shields pooling their regen.

We should've lost this game before the Silo even got up, but the opponents passively let us keep superior map control and invest in Singularities even though they had a large total value advantage. There's a whole nother set of Lessons Learned that could be drawn from that but I've already spent too much time on this analysis.