How deleting/pausing ads affects your QS

[roodle]

Hi,

I'm a relative noob with Adwords and just recently realised that 'deleting' ads doesn't actually really delete them. I'm really curious now about how 'deleted' and paused ads affect the QS of keywords in the adgroup, and general performance of the live ads. Can someone clarify this for me? For example, when would it be necessary to delete an ad as opposed to just pausing it?

Thanks in advance.

[LucidSW]

QS is calculated only on active ads. Obviously, a deleted ad is not active so doesn't figure in the QS.

I also advocate not to delete ads as you may want to activate it again later. Better to pause ads. Also, editing an ad deletes it so don't edit either. If you re-activate the ad, the system has past data to go on for QS instead of starting fresh which is what happens on any newly created ad, even if you re-create a deleted one exactly.

Important to realize that QS is shown at the keyword level but is really a calculation of the ad AND keyword. If you have two active ads, the QS of the keyword is an average of both ads. That's why you should not have more than two active ads at a time as you're just diluting your test.

[roodle]

QS is shown at the keyword level but is really a calculation of the ad AND keyword

I've been running multiple different ads for a long-ish list of keyword phrases in each adgroup. But I'm also beginning to realise that managing/testing the effectiveness of these phrases and ads is impossible this way. How many keyword phrases do you typically have in each adgroup?

[roodle]

That's why you should not have more than two active ads at a time as you're just diluting your test.

I thought with split testing it's only really viable to have multiple copies of a successful ad plus the test ad?

[RhinoFish]

if it stinks, delete it.

[Eschatonic]

I would hold off on deleting keywords or ads rather than simply pausing them, because by doing so you lose past data and the ability to reactivate. The statistics for deleted things get aggregated with no way to separate them AFAIK, so it's usually a good idea to pause by preference.

The only circumstance where I've been forced to delete things is when accidentally running an ad with the wrong domain name (when it was the first and only ad in the group); it had to be deleted before the system would let me use a different domain name within the same adgroup.

I can't immediately think of any other reason why you'd delete rather than pause.

[roodle]

I can't immediately think of any other reason why you'd delete rather than pause.

So if I follow split-testing methods such as this one: [seofaststart.com ] surely I'd end up with hundreds of copies of test ads which won't serve any purpose to be kept?

[LucidSW]

> How many keyword phrases do you typically have in each adgroup?

It's not the quantity, it's the quality. For some reason, many focus on quantity. The more you have however, the more likely you have unrelated keywords.

More than a couple of ads just invalidates your test, especially on low volume.

[Eschatonic]

So if I follow split-testing methods such as this one: [snip] surely I'd end up with hundreds of copies of test ads which won't serve any purpose to be kept?

I believe the maximum is 50 ads per adgroup, per [adwords.google.com...] .

I guess if you're hitting that limit then deleting them is your only option, and that seems like a good reason. Also, if you are losing very little historical data on a particular test ad due to it being a duplicate then there's probably less harm done by deleting it.

[RhinoFish]

you can view past performance data for the things you delete.

[Eschatonic]

you can view past performance data for the things you delete.

How? I was under the impression that it was all aggregated together.

[LucidSW]

> I believe the maximum is 50 ads per adgroup

Correct. But I think that is ACTIVE ads. I never reached that number myself, the most being around 40 ads in total.

> I was under the impression that it was all aggregated together.

You can run and download reports for any period for keywords and ads if you wish and see the historical data, even on deleted data. It's not really deleted.

[roodle]

Thanks for your replies.

So can deleting an ad negatively affect the QS in any way at all? For example an ad which has gained some clicks over time but which doesn't perform as well as others.

[RhinoFish]

i delete lesser performing ads all of the time, in order to increase my qs. i don't mean deleting versus pausing them is magic, point is, active ads should be the good ones.

why do you think deleting a lesser performing ad would harm your qs?

[roodle]

why do you think deleting a lesser performing ad would harm your qs?

Ummmm, maybe because it has a certain value, albeit lesser than that of other ads?

I suppose what I'm really curious about is why Google would keep the ads if they have no importance once deleted.

[LucidSW]

> So can deleting an ad negatively affect the QS in any way at all?

You did not carefully read or understand my post after your original one. Since QS is calculated only on __active__ ads, pausing or deleting an ad will effect the QS positively or negatively depending on the quality of that ad. A poorer quality ad will increase your QS.

Yes, an ad can have a certain future value. That's why I don't delete ads but pause them. But pausing or deleting has the same value as far as QS is concerned. Neither is active.

The reason deleted stuff still exist is simply following standard database practices. A record can be _marked_ as deleted and not be used in calculations. In Adwords' case, you do want to preserve it for historical reasons. Otherwise, you'd come back and say the numbers you run today don't match those you ran last month. The definition of deleted in Adwords is more like "I don't want to ever use it again for my campaign and I don't want to see it". But the data accumulated by that keyword or ad has to be preserved.

[roodle]

Ok I understand now. Thanks LucidSW for clearing this up.