I suggest a different tutorial. This one is rather outdated (for an obsolete version of IMFD) and has a few errors.
This tutorial seems to have you starting on the ground (likely at KSC), then has you set up Target Intercept, then use Surface Launch. This is incorrect for a lunar trip - and is the cause of the "wandering" dV.
One of the confusing aspects of learning IMFD is understanding how Source (Src), Target (Tgt) and Reference (Ref) work. They are related - both the Src and Tgt need to orbit the same Ref.
In an interplanetary flight, such as Earth - Mars, we would set the Src to Earth, Tgt = Mars, and Ref = Sun. We plan the trip using those settings - and the course we plot is actually the course you would use to send the Earth itself to Mars. Then we would use Surface Launch (set to "course") to launch into the ejection plane. Since our vessel is orbiting the Earth (not the Sun), we then use Orbit Eject to perform the transfer burn and leave the Earth's SOI. Once we-'ve left the Earth's SOI (and are now in a solar orbit) we can change the Src in Target Intercept to our vessel (set source to "x", which sets the source to "self") and use Target Intercept for MCC's. Once we are close to Mars, we can use either Planet Approach or Base Approach to make our late MCC's and fine tune our PeA. Then we simply use Orbit Insert to leave the solar orbit and enter a Mars orbit.
It's a bit different with a lunar trip. Our target (the Moon) orbits the Earth - so our source needs to be orbiting the Earth as well. When you are landed, you aren't in orbit. You aren't standing still either - the Earth rotates, etc. What has been causing you trouble is that IMFD is trying to treat your vessel as if it is orbiting the Earth - and the "source orbit" keeps changing.
So, we must launch into a parking orbit BEFORE we can set Target Intercept. We do this by using Surface Launch, with the mode set to "Lunar Off-Plane". For Surface Launch to calculate the correct plane, it needs to know where the Moon will be when we intercept it, so we need a value for TIn. Since the Moon's orbit is fairly slow, we don't need to be perfect here - a day or so either way won't have a significant effect. For a Hohman transfer to the Moon, set the TIn about 4 - 5 days in the future and that will be fine. (For a free return, use 3 - 4 days). Now check the Time to launch, and if it's more than a few hours we will add the TtL to the TIn, and make sure that there is about 4-5 days between TtL and TIn. Once all that is done, we launch when Surface Launch tells us - in the heading indicated - and establish a "parking orbit" of a bit over 200km. Alternatively, we can launch right away (without waiting for TtL = 0) using the direction given as Hed in Surface Launch - it will be slightly less efficient than waiting for a 90 degree launch heading.
Now that we are in orbit, and close to the ejection plane, we can use Target Intercept with Src = "self", Tgt = Moon, and Src = Earth. You can make MCC's when desired - you don't need to wait until you are outside the Earth's SOI (I usually make mine just before leaving the Earth's SOI).
You will find this tutorial more helpful and less outdated, and very comprehensive:
[ame="http://www.orbithangar.com/searchid.php?ID=4142"]IMFD Full Manual/Playbacks[/ame]
NOTE: This tutorial says to use an older version of IMFD for Orbit Insert - however the bug that prompted that has long since been fixed. You can use Orbit Insert in newer versions of IMFD just fine - so no need to keep any 4.x versions. There are also some errors in Target Offsetting and Slingshot sections - we've been meaning to update this but haven't found the time yet.