At the moment the IW goes from start to finish through multiple stages.

At each stage the validation process throws up issues that exist with the data.
So at each stage you have new problems identified that need fixing with the data - so you come out of the wizard, correct the data and go back through the multiple stages before progressing to the next level.
When the dataset is of significant size - this can be quite time consuming.

Its just like a video game.

As the validation process of the IW is so useful in identifying the problems it sees in the data, it would be far better to have a system whereby all the validation elements can be seen at the same time (through a series of tabs) which allows you to surf your dataset, check out all the issues at once, sort them all out in the core dataset, before trying to go through the wizard again.

So each validation section runs in parallel rather than in series.

Does that make sense?