Another good idea is run a lot of keywords on low bids to get the impression results( you will be surprised how high some of your ads still show), You burn some cash this way but you get some important statistics on which keywords ACTUALLY get the traffic, if you do this for long enough and change your bid you can also analyze the difference in impressions VS clicks a higher place gives you,
I started off a campaign this way then upped my bids and paused various sets of keywords for a few days at a time to see which ones converted for me ( by checking the results from the affiliate partner i was sending them to ).
Another approach is to go for long tail phrases and avoid more common generic phrases that are at the base of a keyword tree, for example;
beds - level 1
leather beds - level 2
These phrases are often more expensive as they are the most obvious phrases people decide to bid on and offer less of a conversion rate, as people could be typing these for a number of reasons
Check these estimated type in results next to there average CPC.
Keywords................Avg. CPC........Local Search
[leather beds].............1.38.............18100
[leather bed]...............1.04.............12100
[cheap leather beds]....1.58.............1300
[leather bed for sale]....0.04.............880
[cheap leather bed]......1.21.............320
[leather beds for sale]...0.04............170
[buy leather bed]..........0.04.............16
( for people that don't know on goggles keyword tool in the drop down box marked "Choose columns to be displayed" there is a option to display the average CPC )
Out of these phrases i would be considering
cheap leather beds
leather bed for sale
leather beds for sale
Basically try to think of the phrases that show more intent to purchase.
A user typing in 'leather beds for sale' is much more lightly to buy than one who types 'leather beds'
And balance this with the phrases that have high estimated type in( or using your new data actual type in) and ultimately low estimated CPC
'Cheap leather beds' might be more expensive but it is so for a reason, It converts better than 'leather beds',
The rest of what you do to make that click convert once you have paid for it
Is down to where you land the user and how you warm them up to buy,
It is no use paying for clicks if your sending them to a morgue.
To achieve a better quality score ( higher placement for cheaper or higher places for longer)
You want your advert ( and your landing page ) to relate as closely as possible to the words that the user typed in
So if you are bidding on the keywords 'leather beds' you ideally want leather beds somewhere in
1.In the title of the advert
2.in the advert description
3.the URL,
4.the landing page
( without over killing them )
Goggle wants the user to be served the most relevant adverts in relation to the keywords they type,
and rewards these adverts with a better quality score
And what you want
Is for the user to trust that your ad is the right one for them
And that the place it takes them to does what it says on the advert
O and of course, to make you some money when they get there 
Bidding on 'Exact' search terms in my opinon is better than 'Broad',
you are in exact control of what keywords generate your adverts, you could be paying for loads of phrases that have your keywords in that dont stand a chance of converting a sale.
Try and keep your ad groups tight.It is alot more work but offers better returns.
For example if your were experimenting with loads of keywords try to keep them in no more than 4 to 6 phrases per Ad
You might want to leave these two on there own
cheap leather beds
cheap leather bed
with a advert title of 'cheap leather beds at xxxx.co.uk'
or add them to
leather beds for sale
leather bed for sale
Then Tailor the ad for that group to have all of those keywords in, Maybe a title like
'cheap leather beds for sale'
you have in this phrase
'cheap leather beds' and
' leather beds for sale '
but if you have time i would personally go for as smaller ad groups as possible, if you relate the ads as much as possible to the groups keywords your quality score will be higher and you stand a good chance of converting better
I finally decided that after i had experimented with all the Keywords i could find for the chosen niche, it would be better focusing on 5 or 6 that i had proven to convert.
Then I re-wrote the adverts
for each one of them so that they were all unique and then landed them on parts of the site that were best suited.
Another thing to point out is about placement, i have not experimented much with this but i understand if you know what you are doing you should be able to generate a better return than search. the example i was given was selling ring tones and choosing lyrics sites to put them on
ring tones by the way can be purchased by phone number by ages 16+ , not much else can be bought this way especially by a age group this young
I Hope some of this helps,
But please remember it is just my opinion
