Hey all, Has anyone tried making multiple ads active (like 5 or 6 ads) at the same time in order to let Google optimize them to find the highest CTR? Everyone always teaches that you should only test 2 ads at a time and select "Rotate: show ads evenly." I have been trying this method (let Google Optimize) and have found that it works well. Google seems to "predict" which ads will do better - I believe they do this by comparing your ad to other advertiser's ads that have been successful and begin showing these "winning" ads more often right from the beginning. I am curious why so many people repeat the same teaching - my point is this: Google knows what is going on across the entire network, so why shouldn't we let them optimize ads? Thanks in advance for your feedback - I am wondering if anyone else is using this method with good results. Internet Marketing For Businesses
I guess it is down to the human touch? Google can not read each advert and interpret what a would encourage a human to click however I guess supporting your side you could say Google uses algorithms from millions/billions of similar ad impressions to help make a decision? I've not tried it but may be work testing against a traditional campaign
The Optimize setting can work well, but not always. That's why I recommend the Rotate setting. I would however not test more than two ads at a time, three at the max but only if there is a high number of impressions, like a few hundred per day on each ad. What Optimize does, as you know, is show the ad it believes is the best more often, based on CTR. However, it is a bit too quick sometimes to determine this and that ad gets most of the impressions which naturally skews the results even more to that ad. Use Rotate and decide for yourself once each ad gets at least 10 clicks.
Rotating the ads evenly is definitely the way to go. Don't let Google decide this as it doesn't use conversion/ROI (Return Of Investment) A data which is much more important than CTR. With regards to what Lucid said about leaving it 10 clicks. I'd suggest at least 30 clicks is more what you need. Split test A vs B and make sure the better performing ad has had at least 30 clicks before you delete the less performing Ad. If the statistics for both ads are too close to call after 30 clicks, then leave them until 40 or 50 clicks.
Yes, You make a great point: just because an ad has the best click-thru rate, does not mean that it has the best conversion rate. I remember a few years ago using ads with "FREE" in the title - got really great CTRs, but conversions were not always great. I still wonder if we are using conversion tracking, and really let Google optimize, would they factor in conversion data in order to determine a winning ad? I am still thinking that this "let Google Optimize" feature could be a hidden gem. I found it to be especially helpful when running a lot of adgoups - Google will do the job of picking a great ad while I am away from my computer- the reality is that I don't always have time to monitor and mess around with pausing and deleting ads every day. I think that given enough time, Google will find the best performing ads (best CTRs that is) and then you could go back later and make decisions about ROI. I really think that we all have been sort of brain-washed by the repeated teaching that rotate ads evenly is the better option. I am going to do some more testing, still interested in hearing back real-world results from the rest of the group. Thanks! Business Marketing Speed Up Your PC
A hidden gem? It's certainly not that. My advice: if you are serious about improving your Adwords account, *do not* use it and select to rotate the ads evenly. You can't split test properly if the ads aren't being rotated evenly. And if you aren't split testing... well, more fool you.
Muchacho, Maybe you should try the other option. It sounds like you have never really tried "let Google Optimize" I am interested in hearing real results from those who have tried it, not arguing with those who have not. Internet Training
> You can't split test properly if the ads aren't being rotated evenly. That's right. You want a fair test and the only way to do it is an even number of impressions at all times. You don't want an ad showing more often in the evening for example, this would skew the data. What if Google thinks the best ad is A just because it ran more often at a certain time of day or day of week just because those were the times that happened to be the best for you? Ad B would be declared the loser but that would not be a fair test as it might be shown more often during the times when both ads would be equally crappy. Statistically, 30 clicks would be better than ten. Ten is fine if you have 250 impressions or less. I think the Optimizer uses around 30 impressions, at least I recall seeing that number somewhere, and that's not enough.
I've tried it before. It didn't take conversions into consideration. Currently, Google aren't better than us humans at deciding which is the better Ad when Split Testing. It's that simple. The only way people would ask Google to decide the better ad, would be if they'd never Split Tested before and just assumed this would be the best option.
The 30 clicks with Split Testing is something I read about a while back and this figure was supposedly mathematically correct, but can't remember the reason why. I'm not sure why 10 clicks would be fine if the impressions were less than 250, you'll have to explain that one A lot of people see CTR as important and really, when you think about it - it's not. With my split tests, I'd choose A over B if it had a better ROI/Conversion data but a much worse CTR. I notice a some people automatically go for the CTR. I think the reason why they do this is because of the way Google put it across. CTR helps QS but people don't realise that Ads also have their own Quality Score. If you test A vs B, Ad A might get a better CTR, but the reason for this *may* be because Google likes it more than Ad B and gives it a higher Quality Score, so it's ranked higher. A higher ranking will more often than not mean a higher CTR. People need to see past the CTR and look a lot harder at the ROI/Conversion data. That's what matters. I'll often put 'filters' in place to hopefully prevent some people clicking my Ad, such as mentioning a price, or an age range (over 18s, for example). Now this may mean the keyword Quality Score will be less than it could be, because the CTR drops, but like I say, who cares? It's the Return of Investment that's important, not the CTR.
> The 30 clicks with Split Testing is something I read about a while back and this figure was supposedly mathematically correct, but can't remember the reason why. Something to do with the point at which each successive click would not improve the accuracy and within a tolerable margin of error. It's like polls where they say "accurate to 95%, 19 times out of 20" or something to that effect. Say you have 30 clicks in 750 impressions (4% CTR). One less click or one more only changes the CTR by a small percentage, about 3% either way, which is within the tolerable error. Same thing when you reach a certain number of impressions such as the 250 I mentioned. If I have at least ten clicks in that number of impressions or less, that is statistically significant and within a tolerable margin: waiting one more click or ten more impressions is not going to make much of a difference, especially if one ad is clearly ahead. But you're right, best to use at least 30, not ten, although I know of a company that uses a minimum of ten clicks to decide a winner but it may be only because they don't want to wait too long. > A lot of people see CTR as important and really, when you think about it - it's not. With my split tests, I'd choose A over B if it had a better ROI/Conversion data but a much worse CTR. I do the same. Actually, I multiply CTR by conversion rate. Higher one wins. But here too you need to have enough conversions to make it statistically significant.