Inconsistency in AdWords Keyword Tool?

Weak coverage of single-term combinations

[deus]

We found some evidence for an inconsistent behaviour of the new keyword tool at
[adwords.google.com...]

See my examples:

Settings: all languages/all countries/no synonyms

1. parkinson disease symptoms

   -> 6 combinations
   -> okay.

2. parkinson disease

   -> 201 (max) combinations
   (including "parkinson disease symptoms") -> okay.

3. parkinson

   -> 23 combinations only!
   (NOT including "parkinson disease") -> missing data - should be equal or more combinations than example 2.
   

So, doing a solid and consistent keyword research is absolutely essential for a successful AdWords campaign - you all know. If we can't go from the general terms (e.g. "parkinson") to the more specific terms (e.g. "parkinson disease") we have a big problem.

[mike_ppc]

Usually G's estimators were very far from reality. Just remember Traffic estimator - what huge traffic it would predict!
You can use the estimator only to have an idea, not to depend on it.

[deus]

Mike, I'm not talking about traffic-estimator numbers here - at this stage it's about discovery of possibly relevant search terms. Sure, interpretation of the tool results is one thing but internal consistency of a tool another. I'm sure the appropriate data is there - the former keyword sandbox did a great job at this point.

[mike_ppc]

The idea was that you usually cannot trust an estimator.
Sandbox is useful because it provides interesting keywords, but you DON'T depend on it.
If you use your Traffic estimator to build a budget for a client and you see later that there is a big difference in reality, what can you tell that client? That it's not your fault? Because you really DO depend on that estimator for some purposes...