Hi, I'm going down the route of trying several different adverts to guage clickthrough and conversions and was just wondering how the pro's on here work there's? Do you pause the old advert and run the new one for a week/month or what? How long do you test the new advert before you tweak it some more? Thanks
I usually create 5-10 and let them run for a few weeks then remove the lower performing ones and stick with the top 1-3. Of course if one is doing really bad over the course of the week I remove that asap
I generally run either two or four adverts. Usually, it's two, with just the one change, but if I think the lines may be interacting, then I'll try all four combinations of two changes. For example, Advert 1: Buy Stuff It's Good It's Cheap Advert 2: Buy Stuff It's Good Free P&P Advert 3: Buy Stuff Lots Of Colours It's Cheap Advert 4: Buy Stuff Lots Of Colours Free P&P In terms of how long to let them run, I wrote a blog about that: http://www.epiphanysolutions.co.uk/...g-is-statistical-significance-over-rated.html I also discuss the whole question of how to test adverts here: http://www.epiphanysolutions.co.uk/google-adwords/multivariate-testing-on-ppc-adverts.html Enjoy!
I thought most people run each set of advt depending on the CTR. Say first 1000 or 5000 impressions, and then dice and slice.
Run two adverts intially...track theor performnace ater a week or 10 days...keep the better performing ad, and replace the poor one with some modificstions done keepping the first one in mind....
great example custardmite i would definitely run more than one advert and see which one performs better. Dont just look at CTR also look at your conversion rate
Of course, it depends on how much traffic you're getting! One of my campaigns get enough traffic for me to test a new advert every two days, but another one only gets a fraction of that, and I can only try a new advert about once per month. Also, the size of the difference between the two adverts' performance matters. If one advert is MUCH better than the other, you can kill the worse one much more quickly, as the difference becomes apparent much sooner. What I would say is that it's not worth waiting for statistical significance, if the performance of the two adverts is very similar. Just get rid of the worse one - even if it's better, it's clearly not much better - and try something new. It's better to try ten new adverts, and make the right choice nine times than to try three new adverts, and get the better advert every time.
Yes, Absolutely right, but test the advert time to time because, according to customer need and to attract new visitors you need to test the adverts. It updates you with market.
Damn looks like I'm going to be writing a hell of a lot more adverts then :-< Thanks guys for all the good advice, I particularly like the bit from CM about getting rid of the worse even if they're close - sometimes a decision has to be made dammit!
Run until www.splittester.com says you're at least about 85% sure. If it says you're not sure, keep 'em running!