How are you creating your list of keywords?

[dylan212]

Hi

I am curious the various ways people create their list of keywords used in ppc engines. I realize some people have 10,000+ keywords in Adwords but not sure the best ways to generate a list this large.

Let's say for example I am using ppc to advertise 'Mortages' and I have a site which generates leads for mortgage companies (i would say 'widget' or something but i think a real world keyword and business would be more helpful here - btw, i don't have anyting to do with mortgages).

There are so many variations for advertising 'mortgages' (ex. refinance mortgage, first time mortgage, etc...) but can anyone offer suggestions, tips, software which will help generate a comprehensive list of keywords I can use for Adwords? I'm prepared for the long hours and days developing these lists but some experienced knowledge here could save me, and others, some time and frustration.

Cheers

[running scared]

Hi Dylan212,

It's a good question and an area that I would love to work out a nice easy system.

My main tip is to make lots of use of excel to manage keywords and you can use functions to generate phrases by joining up words.

Our process looks something like this:

1. Review site looking for key phrases
2. Review competitors sites looking for key phrases
3. Brainstorm with colleagues
4. Use Overture keyword tool
5. Use Adwords keyword tool
6. Use Wordtracker
7. Panic about how to organise all the output :)
8. Continually add new phrases as campaign continues.

(side note - we try to avoid talking about specific industries at WebmasterWorld, hence lots of chat about widgets on the forum - chances are someone here is working the mortgage space)

[John]

Run any analyser on your web logs and then strip out the actual key words your visitors have used. This is a source that actually comes from the minds of the public. What will appear may look odd, wacky and even plain stupid but it is what the public actually puts in. My catch all adgroup containing this almost junk like list outperforms my "rationally" created list.

[dylan212]

Thanks 'John' and 'running scared'

I agree with both of you and think they are all very helpful tips. I have some further questions...

When creating a list do you also add common words to all your keywords? Let's say for example (and i will use the 'widget' scenario from now on - thanks running scared), I have the following keywords:

red widget
blue widget
green widget

Would I want to add the following words/phrases to each of the above 3 keywords:

looking for
searching for
how can i find
buying
i want to find

Thus:

looking for red widget
searching for red widget
how can i find red widget
buying red widget
i want to find red widget

looking for blue widget
searching for blue widget
how can i find blue widget
buying blue widget
i want to find blue widget

looking for green widget
searching for green widget
how can i find green widget
buying green widget
i want to find green widget

I imagine I could come up with hundreds of phrases like this.

Further, how do I decide whether to use Exact or Phrase matching? I understand the differences but not so sure how I determine which one I want to use.

Whew! Hope that was simple to understand

[John]

I have done exactly that and find I continue to mop up more an more marginal searches. These 3,4 ,5 and even 6 word searches convert as well as or better than the obvious ones.

I think some people put in 2 or 3 words and then add more when the results don't meet their needs. They don't necessarily add new words on logical order hence the stage phrases that still pull results.

For my red wedgets I have a full listing of 7500 plus key words. I am currently building a tool for cranking these out but it is early days!

[dylan212]

thanks john!

how are you determining exact vs. phrase match? do you use all 7,500+ keywords for 'red widget' as exact or phrase match? can see you differences in impressions, ctr, cpc when comparing the two matching options?

i'm not sure why or when to use each matching option. i have noticed some people use both matching options in the same ad group.

"looking for red widget"
[looking for red widget]

[ogletree]

I use MS Access to make kw lists. It is real easy when you have a list of adjectives. Make one table that says color and has red, blue, green etc... Make another table named cost that has cheap, inexpensive, low cost. You can find out combos at Wordtracker to help you make these lists. Make a new query in Access and in the field put

Expr1: [cost] & " " & [color]& " " & "widget"

run the sucker and you have a large list of words. There are many variations of this.

[John]

I used the broad and exact match usin "" and [] but when I also run the manually produced BIG, list with every permutaion and combination in, I get a lot more clicks and impressions. I don't know why this is.
o
Maybe if you are broad matching and so is everyone else you are sorted by some other criteria. Maybe if you actually have the exact phrase in your monster list you get extra brownie points. Who knows?

If you crunch out the list you are not at the mercy of a google algo or some other syndicated engine.

I also use the technique of listing lots of possible adjectives. I also make sure that I have all the " learning, learn teach, teaching, finding, getting rid of, get rid of, find, finding...etc,etc" They all produce productive clicks.

They key is them to load em up, run for a few hours and look out for any goofs where you are burning up money. These usually turn out to be acronyms or names of songs or bands. Kill em off.

Them let the adgroups settle. Slice and dice the good terms off to new groups and polish the ads.

Good impressions but no clicks get skimmed off to a new ad group and get a radically new ads and the non performers for impressions get moth balled.