I've just added a paramater to one of my Ad Group's Ads destination URL, but instead of just editing the ad - it's deleted the ad before adding the new one. Why can't it just edit the ad and keep the stats going?
The changed destination could potentially change the quality of the advert, if the landing page is now different. This means it is effectively a new advert and when comparing it to another for split testing the results are not fair. It may still be the same landing page but Google won't know for sure - perhaps they will bring out an option to edit wthout deleting in the future.
Prodrock hit the nail on the head. When wanting to alter a variable within an ad, just pause the old ad and make a new one with the desired change. If you want to test a new ad, the best thing to do is to make 3 extra copies of the old ad and one copy of the new ad, so that if the new ad underperformes, your overal adgroup performance will not be affected. Never edit ads. Just create new ones.
"If you want to test a new ad, the best thing to do is to make 3 extra copies of the old ad and one copy of the new ad, so that if the new ad underperformes, your overal adgroup performance will not be affected" The thing is, certainly in my experience - pausing an ad, is just like deleting it, in Googles eyes - they don't take it into account anymore when working out QS - if you unpause it, Google starts taking it into account again. If you mean make 3 copies and have them ALL online .. that would surely defeat the point of a split test, as it wouldn't be a fair ratio.
Pausing an ad is different in the respect that it retains performance data. This speaks directly to the TS's question. When comparing the three old 'ad clones' you have to aggregate them in relation to the new 'stand alone ad'. This is a known method of testing ads.
OK I think I get you - so it's basically a way to work around the fact that Google for some reason make it a new ad when all I have done is change the destination page slightly, by adding a parameter on the end. Everything else is the same. So if I have Ad1 and Ad2 and I'm changing the parameters on the end of each. I pause them both and then make exact copies just with the parameter difference - in a weeks time when I'm comparing Ad1 to Ad2 I ad the Ad1 data to the one that I've paused, to work out the figures - same with Ad2. I still don't know why one would copy it 3 times though. Copying it once, yes - as you have older data to add to the newer ad, but 3 times?
If you copy an ad three times, there will be 4 version in the adgroup (3 new + 1 original). Quite often the new ad you're testing will NOT beat the original one and if you run them head to head (1 0n 1) there's a good chance that the adgroup will be dragged down. By creating 3 new versions of the old ad you cushion the blow in case the new ad is bad. If the new ad is better than the original 4 (which are all clones) then you just turn them off when you're sure the results are not due to chance.
Ok thanks for that, it's not something I would consider if I hadn't read this. But ... what about the fact that the ratio not being fair? If a new ad is only going to be shown 1/5 of the time .. with the other 4 being displayed 4/5 of the time. Surely this isn't a fair split test?
Well it's a question of what metrics you're considering when determining which ad is best. Usually it's going to be base 100 (percentages) with CTR and Conversion Rates. The only thing you have to aggregate are the four CTRs and Conversion Rates of the four clone ads. The actual number of impressions and clicks is irrelevant. If you're using CTR as the key metric in determining which ad is best, you make sure you take an average CTR when comparing the four clones to the lone new ad. Same for conversion rates. As long as you have enough data, this will work and you won't lose adgroup performance.
the best way to preserve ad copy is to not delete the text ad, or manipulate it in any way. So to do this, set the destination URL at the keyword level only... Problem solved