I recently did exactly what you're talking about, and RFS was awesome for the job: its open structure let me model many different play strategies and styles quickly and easily just by applying different philosophies to different actions and scenes.
First, the big pitfall I encountered: Roll For Shoes doesn't do much heavy lifting. The burden to make the game work falls entirely on its participants. As a teaching tool the system creates many opportunities for learning, but it's the responsibility of the GM and players to recognise and exploit them. Following are some opportunities I took advantage of.
Talk about choices.
I attribute my success mostly to frequently pausing the action to discuss what was going on in terms of playstyle choices. Instead of expecting the player to pick stuff up through osmosis, we had explicit conversations about the choices I was making, the choices I was asking him to make, and how they influenced the game experience. This also helped establish the collaborative environment I've come to value in my groups, blurring the lines between the social roles of player and GM at the table.
Trade off narrative duties.
Roll For Shoes has no interest in dictating who's responsible for determining outcomes. I'd frequently ask the player to describe what he learnt from a successful roll or how he failed his action on an unsuccessful roll: when the ninjas knocked him out, I asked him where he woke up and who was nearby. Similarly sometimes I'd tell him the cost of success, and sometimes he'd tell me.
Zoom in, zoom out.
We started with a "traditional" combat scene with turns and rounds: one character rolled to take cover, then the next PC rolled to attack, and so on. Each roll corresponded to one granular action.
Two scenes later, I told my player that we'd zoom out for a kind of "montage" effect: we used a small handful of rolls to describe the entire process of PC's infiltrating a space station. Each roll corresponded to multiple actions over minutes or hours of time.
Games like D&D can do this to some extent with skills, but can't handle combat on anything except a turn/round basis; Fate's contests and challenges are designed for zooming out the action, but it's got a pretty rigid structure for it. Roll For Shoes just lets you wave your hands and make a roll's effect be as narrow or wide as seems right, so the new player can see how this is done before learning a specific structure for it.
Roll to act, roll to discover, roll to declare.
Roll For Shoes is at its best when you're rolling for everything. If the player asks whether there's a koi pond in the zen garden, or says he wants his character to have a space ship, or has his character try to catch a shuriken and throw it back at the ninja, I call for a roll. This provides speedy growth of skill trees, and also gives lots of opportunities for learning stuff like...
Failing forward and succeeding at cost.
When the player first failed a roll, I talked about failing forward: the idea that failure should never stop the story, but instead reveal alternate avenues of action. This made him less afraid of being defeated by the attacking ninjas, because he knew defeat wouldn't bring the game to a halt.
(RFS helps reinforce a positive attitude toward failure through its XP mechanic, which I pointed out to the player: failure provides the building blocks for future growth. This is a role-playing lesson I can't emphasise enough, so it pleases me that RFS supports it mechanically.)
And when he failed at a roll for something he really wanted to have happen, I introduced success at cost: a failed roll doesn't have to mean you failed at the action, but instead that some new complication is introduced in addition to your success. Thus, while the PC was sneaking onto the space station his inside accomplice got caught and the station security went on high alert.
Success at cost (a phrase I'm borrowing from Fate) isn't baked into the Roll For Shoes rules, but is the result of a liberal interpretation of "the thing you wanted to happen, happens" in the second bullet point. Your mileage may vary on whether you're comfortable using this approach, but it's been very useful for me--so long as the costs have bite.
