When to switch off AdWords ads in your channel mix?

[Cesare.Marchetti]

My question is actually Inspired by that very good article: https://moz.com/blog/google-organic-clicks-shifting-to-paid that I read this morning.

Present situation:

For a specific and valuable non-branded keyword combination (2-3 words) we rank:

• Google Adwords:  position 1

• Local Pack (with maps): position 1

• Organic Search: position 3-5, lately more 5-6

Question?:

• Is it still worth having AdWords ads or not there?
• How to evaluate if we could do without AdWords?
• Any algorithm, experiment, thought that would help find out?

We know our average cost per lead for the different channels. Just to leave away AdWords ads for a certain time is not really an option nor would that statistically mean anything, i.e. if we skip AdWords and have the same number of leads after that and compare it with the months before (with AdWords) that could also be due to other reasons (seasonal aspects, etc.).

Put in other words: if we skip AdWords people would still click (more) on our other two search results (local & organic). I am not sure if the additional leads coming from AdWords outweigh the cost we have for that.

Would love to know your thoughts about that.

Thanks a lot for your input in advance.

Cheers,

Cesare

[Alick300]

Hi Cesare,

Adwords top three ads appears above organic results, even if you are on top in organic for the queries,user may get attracted with adwords ads as they have proper call to action, sometimes prices sometimes offers and also sitelinks, call extensions etc.

So IMO you shouldn't turn off adwords campaign even if you are ranking on top position in SERP.

More times your domain will appear more it can catch users eyeballs.if user see ad from single domain more then once they are more interested to click on your ad.

Run a campaign for a week, turn it off for another week, then on again and off again and see if you can detect any pattern.

Thanks

[Cesare.Marchetti]

Hi Alick,

Thanks for your answer. The more probabilities to catch the user's eyeball the better. Right. But with AdWords its also going to cost something...lol

I know of people that would never click on an ad. Me included, AdWords ads excluded actually because I think that the often are really relevant.

Run a campaign for a week, turn if off for another week and then trying to detect a pattern is actually not the statistical proper way to do it. What you see can be biased by different things that happen exactly during that week. An A/B (split) test would be the appropriate way to do it, but honestly I wouldn't know how to set that up in such a context.

[Moreleads]

Hi  Cesare Marchetti,

Thanks for you question.  I think there are several things to consider here:

1. are you the only one appearing on adwords for these keywords or are there other competitors as well.  If latter is the case you would simply be loosing potential business to your competitors as 20 - 30% of the users click on adwords (these are the ones left over after the 70% - 80% that never click on ads) for commercial terms.
2. Different people with different intent, devices, place etc.. will click on different results so i would like to dominate the serp as much as possible so i can cater to those, and adwords would just be one of those results.  This way i think i can maximize the results and return.
3. But these are mostly empirical assumptions to really test this objectively is almost impossible as there are many factores influencing results.  like you said seasonality, competitores, positions you have in the serp based on device, location, time etc... You could stop adverting adwords for one month for one of the keywords, compare how results were for this keyword compared the other keywords (and compare that to how they would compare in previous months), ajust for seasonality based on historic monthly results. But there can still be other factors incluencing the results.

Hope this helps and give you some perspectives on the issue, althouhg i don´t think this is completely what you were looking for (and if somebody does know a way to test this i would be very interested)