when to kill a low CTR keyword

[hudson]

Hi Guys, maybe this has already been addressed, but my question is when to kill a low CTR keyword...what is a statistically significant sample in other words?

100...200...500 clicks?

I'm thinking a thousand, but then testing three keywords that don't work out would cost $5 (to reactivate the account three times due to low CTR).

[eWhisper]

I don't know what's considered statistically significent, as some words get thousands of searches a day and others can generate only 2-3.

Personally, if I've chosen to advertise for a keyword, then if I'm not getting a good CTR, I try to figure out why and improve it's rate. I 'back' my words up by not just deleting them because they didn't preform well at first, I first assume I didn't write a good enough creative, write several more, or didn't target it correctly, and look at negative keywords to add and ways of making the word get a better CTR.

[Robsp]

hudson,

It is also smart to 1st run your campaign for a while with the better performing keywords before you start experimenting with new or low CTR ones. This saves you the reactivation fee. You can first delete keywords and get them back in later once you have a decent base CTR for your full campaign.

[hudson]

Thanks for the reply, guys. The problem I'm running into is say you got a bunch of widgets you want to sell:

Acme Widget
BigCorp Widget
SmallCorp Widget
Fancy Widget
Budget Widget

So, naturally, you are going to try:

"Acme"
"BigCorp"
"SmallCorp"
"Fancy"
"Budget"

With results in a few hours appearing...some work like a charm, others are too broad and maybe goes over the 0.05 mark.

Well, I am still new to this. I appreciate the suggestions to start with a core group and work on getting a better CTR on terms instead of throwing them away.

On the other hand, I'm still curious if something that gets 250 impressions without a click might turn into a winner if you give it time. Or if 40 impressions without a click is a mark of something you should pull (or at least rethink).

[eWhisper]

First of all, if "acme" sells 1k products, and you're only selling 1 of their products, I don't think that's very targeted, and while you might end up with a high CTR rate, odds are your conversion will be low.

If I think it's a good keyword, I let it run for at least 5k impressions (if it's getting over 0.5% CTR) or 2-4 weeks before I yank it. If I'm not sure it's a good KW, then I monitor it very closely, make several ads to see if one seems to work well, and check on it's ROI factor at the end of each week to see how it preforms.

There are too many variables that can make your ad appear a lot of times in a short time span which don't gather enough clicks to really determine it's quality.

Have you ever clicked through 10-30 pages of serps? That would generate 30 impressions right there, and if that person never clicked on it, you might think it's not worthwhile, yet only 1 person even viewed the ad.

[hudson]

Hey, thanks for the reply eWhisper. Read "acme" not as a company, but as a brand name. But, from your reply, I gather if a kw is not generating at least 0.05% it is not worth running with.

Have you ever clicked through 10-30 pages of serps? That would generate 30 impressions right there, and if that person never clicked on it, you might think it's not worthwhile, yet only 1 person even viewed the ad.

Yes...I noticed this! The drawbacks of raw impressions, I guess. It made me think if you get on a SERP with only one or two bids and people are searching, searching, searching for something that could cause some problems for CTR.

[AdWordsAdvisor]

...say you got a bunch of widgets you want to sell:

Acme Widget
BigCorp Widget
SmallCorp Widget
Fancy Widget
Budget Widget

So, naturally, you are going to try:

"Acme"
"BigCorp"
"SmallCorp"
"Fancy"
"Budget"

This would seem to make sense at first blush, but really those keywords are way too general, and will most likely get poor results. Especially 'budget' and 'fancy', because those words actually have very little to do with your product - which is widgets.

IMO, if have 'budget widgets' for sale, you would want to run on the keywords such 'budget widgets' - and so forth.

I strongly suggest using keywords that exactly describe what you have to offer. That way only people who are looking for exactly what you have will see your ad.

This will likely bring you a pre-qualified customer, as opposed to the window-shoppers you might get with very general keywords.

AWA

[skibum]

Sometimes it can be to ones advantage to run some of those general keywords. If you can capture the actual keyword used to search instead of just what was purchased, you'll be able to come up with a list of negative keywords and come up with very highly targeted ads on the money terms.

Sometimes the best terms to bid on are ones you'd never think of. People think of and view products and services and express thier needs and wants in all kinds of strange ways.

As far as minimum impressions, it might depend on the goal of the campaign but generally ROI is more important than CTR. Would ya rather have an ad that gets a 5% CTR and doesn't convert or one that flirts with the minimum CTR to stay afloat and rings up a sale every time?

500-1000 impressions can usually provide a good indicator. Creative changes can sometime work wonders too. Not always the keyword that is to blame.

[hudson]

Thanks skibum for the outside-the-box thoughts. Also, I think you are right that 500-1000 would be a good cut-off point.

And AWA, I appreciate the idea of not going too broad to avoid tire kickers.

This whole AdWords thing is starting to seem more like an art to master ;-)

[danieljean]

Says Jakob Nielsen, "there would be a less than 3% probability of experiencing 355 exposures without a single click. This is more than enough proof. Of course, there is a 3% chance of canning an ad that might ultimately work, but so what? Better to try something else rather than continue an experiment that would be a waste of money in 97% of the cases."

HTH