Forum Moderators: open
heini is absolutely right. 50 million searches should be considered a bare minimum if more than one language is used and any engine anywhere will have searches in local language + English at the very least. Even 100 million is not too much.
If not, you don't have to take the numbers as Gospel but if they show that most people search for KeywordA rather than KeywordB then it gives you a good idea of what to go for.
I wouldn't just dismiss a keyword tool out of sight - anyway, 50 million sounds enough of a base to start from.
Having thought some more, it occurs to me that there is a simple and nowadays also effective way for data mining at low cost:
1. Set up an Adwords account with Google.
2. Look up the Broad Matching option.
3. Type in every single keyword and phrase you can think of.
4. Examine Google's suggestions for additional matches and use them as appropriate.
5. Use their tool to calculate clicks and costs. This will give you a first rough idea.
6. Set click cost to the minimum or whatever you can afford.
7. Let the campaign(s) run for a week and check the figure for Impressions. Multiply by 52.
This should give you something almost as good as a full blown database, in a short time and at minimal cost.