when to start eliminating non-producing keywords in PPC

[brizad]

For one of my sites I have about 300 keywords running in PPC. After a few months only a handful have produced positive ROI. Lots of people say that it is important to have many keywords but I am starting to wonder if this is really the case.

Case in point here are some rough numbers: Since Sept $1000 spent in PPC. The 20 or so words with positive ROI only cost about $100, so the other $900 was (in theory) wasted.

Now that I know what my money words are, I could just concentrate my $$ there and delete the others. Right?

On the other hand, I guess I could try to keep tweaking the underperformers to see if I can get them into postive territory.

Any thoughts on when to pull the plug? I've heard Michael Anthony say 100 clicks/3 days/$100 but I am not sure if this was referring to a campaign in general or specific KWs.

Thanks...

[wellzy]

You have given them way more time then I would of. I would cut the non-producers off now. I give them a few weeks or 100 clicks. Time to try other keywords.

Keywords that are not producing well will not start to produce eventually. This is based on all of my trial and error with keywords.

wellzy

[moneymancn]

We eliminate high cost,non producing Kws immediately.
Low cost ,non producing have a maximum life span of 14 days as do "No clicks" Kws

We do rotate these back into the system to see if it was a seasonal,or other, reason for no clicks though.

MM

[brizad]

Thanks for your replies. This sounds logical to me too. I was listening to the advice of some so-called "marketing experts" who said you should try hundreds if not thousands of keywords to expand your reach and the possibility of clicks.

Low cost ,non producing have a maximum life span of 14 days as do "No clicks" Kws

Do "no-click" KWs hurt anything? I have left them there thinking that they weren't hurting anything and that they might get some clicks eventually.

Wordtracker, overture's KW tool, etc. Say these words are searched for so I figured that maybe my ads weren't good enough and that maybe I should keep tweaking them until they produce.

[Michael Anthony]

As a rule of thumb, I try to think up 20 new keywords for every one I delete. It's essential to find those low cost keywords that nobody else thought of that really convert - let G Adwords disable the losers, delete the ones that don't convert, but keep thinking - plurals, mispellings, competitor site keywords metatags, competitor site's copy, just keep piling them in.

It's important to have lots of keywords, and you only find the nuggets by constantly mining - you'd be amazed at the number of sponsored link listings you can achieve at very low CPC's just by adding and adding to your keyword list.

One competitor of mine in the loans sector has an entire page SEO'd for the mispelling "laons" - bet he makes a killing. And I wish I'd have thought of it first!

And my 3 days/100 clicks/$100 rule applies regardless of product or site type. Maybe it's a silly and illogical rule, but it's saved me fortunes over the years and I've never regretted canning a non-performer in nearly three years of doing this stuff.

[tml89]

Anyone use tools to generate mispellings and plurals?

[fclark]

Michael Anthony,
Do you find that your adgroup gets placed into "waiting for evaluation" mode each time you add keywords? Or do you just plop them into new adgroups?

[brizad]

Anyone use tools to generate mispellings and plurals?

W*rdtracker

[PeteM]

Anyone use tools to generate mispellings and plurals?

I don't normally divulge tricks of the trade on open forums but for plurals I normally just add an 's' on the end. Very occasionally I replace the trailing 'y' with 'es'.

;)

[Michael Anthony]

To anwer the post above, maybe some of my new keywords get put into waiting status or similar, but I never bother to look at this level of detail.

My reasoning with Adwords is that if they delete or suspend something because it's not working or they think it might not work, so be it. It's the winners that interest me, not the also-rans.