Slicing Software matters in 3D Printing - Cura vs Slic3r

NOTE: Post Updated (5 weeks later): I've found parameters to make CURA slicing very consistently successful - mostly thanks to the folks at LulzBot who sell great filament and provide great filament profile/config files for CURA - but also after a bunch of experimentation. I've been using CURA exclusively now and while I tweak speed and top/bottom layers on different models for different needs, I have not had a fail like the one pictured in this post since writing the original post... but leaving the post here for historical purposes.
Original post follows:

I discovered that Lulzbot seems to suggest CURA as their prefered Slicing app for generating gcode files for 3DPrinting (gcode is the language that many #3dPrinters understand). I was initially using Slic3r because that's what the Lulzbot TAZ4 manual pointed me towards. Since I'm just starting to understand the basic controls I have in Slic3r, getting a taste of CURA was a great experience - to see how it exposed and set different parameter versus the way Slic3r did.

CURA model on the left, SLIC3R on the right.

At first, I thought I had discovered gold - since one of my models generated GCode much (much!) faster than Slic3r and then the generated gcode printed extremely fast. My second try also showed improvement - printing the CURA gcode in almost half the time on a small model - and the quality was quite good.

Then I tried a new and very different model. Still very simple - some geometric shapes extruding up off a rounded courder rectangle. Sure - it happened to be the Google Classroom logo - but it's a simple one. This is where CURA not only fails me so far, but perplexes me. The top layer is practically not completed - and there is actually a shape missing. Check out the image here comparing the CURA print to the Slic3r print. See the missing shape? (it's a chalkboard eraser, by the way). The SLIC3R model is by no means perfect - I have some work to do figuring out why the top layer has gaps - but it's still much better than what CURA produced.

I still have lots (LOTS) of experimenting to do - but for now, I can see that I might have to start digging into the gcode to see what's going wrong.