Numbers are not yet known for the 2nd stage retrieval or Falcon Heavy. One could probably assume a similar number for the first stage but possibly a bit less given the Falcon Heavy's higher thrust/weight ratio.
An enormous amount of inertia to overcome so 30% doesn't sound so unlikely.
1. Alter the path so the parabolic arc goes back to the launch site.
2. Braking burn during descent to minimize atmospheric stress.
3. Landing burn (AKA "suicide burn") to slow the vehicle enough for the landing legs to absorb the impact.
Though it's not quite that simple. Based on the trajectory of last week's launch, they're actually going to use a much steeper trajectory during the first stage launch. The result is that the first stage uses significantly less propellant to make that first burn (since its downrange velocity is much smaller), but it cuts into the payload capacity quite a bit by forcing the second stage to do more work.
(I didn't know that)
The second stage will need a re-entry heat shield, but otherwise can land like the first. Tricky part is getting enough margin in the second stage for the landing fuel and legs and heat shield. Usually the top stage is the one you optimize most heavily. So plausible, but tricky.
Good news is that the second stage is much cheaper than the first, at least on the F9. Eight fewer engines, for one.
Edit: Found the original copy on the Web Archive http://web.archive.org/web/20110615000000*/http://www.spacex...