Besides organization, the other IA element we're concentrating on is labelling - the specific words we use in our headings.

When we run a tree test, we are seeing the interaction of these two factors:

Organization	If a user can't navigate down to the right heading, it doesn't matter how hard we worked to make that heading clear.
Labelling	If a user doesn't understand a certain heading, they're unlikely to click on it to see its subheadings

Some labels are dead easy to create, while others seem to get harder the more we tinker with them. What is it that makes one label better than another?

Below are some principles and tips to help you create effective headings.

Speak the user's language

need content

Make headings clear

e.g. bus website

Make headings distinguishable

e.g. bus website

Use specific, concrete terms

example from MoE or MfE?

Make headings scannable

i.e. front-loading phrases (see Nielsen article - same idea as for links)

Balance brevity with clarity

need content

Combine entangled topics

(e.g. foobar & bleem)

For more on this, check out ~web or book link.

Next: Team-sourcing ideas