Alex W.
Bronze Contributor
User Power
Value/Post Ratio
164%
- Sep 30, 2015
- 69
- 113
- 31
Going to add this in too...
Even though this isn't "PPC", it is highly related -> Split Testing.
I don't like traditional A/B split testing.
Most people hear and think about this and dump it into 50/50 tests, which means 2 version of an ad or LP running at 50% to each.
Bullshit.
You're not always going to make a "better" or improved version of what you're testing.
Throw up a B version and have it fail, and you just lost potentially 50% of your leads/sales.
Now, there is software like Optimizely for LPs that will let you set the percentage to A as maybe 80% and the percentage of B as 20% which is a much safer approach ( if you know that A converts well ).
Another way to look at this is A/A/B testing where you have 50% go to the first A, and then the left over 50% is split between A and B ( effectively making a 75/25 split). The only reason I bring this up is some software out there will not let you do an 80/20 like Optimizely as there is no percentage input for it. For these types of software/methods, you are just adding in A again.
But this isn't just for landing pages, you can apply this to your ads as well.
No sense in making ads that might cost you impressions and clicks if they are bad ads ( just like my bad LP example above ).
When you are putting in ads, stick to not wasting money on inserting 10-20 ads at a time that are all different if you already have a winner.
I believe Adwords campaign experiments (ACE) allow you to do the same in regard to A/B testing (no A/B/n testing, yet). I don't have direct experience, though, as I'm currently using 3rd party tracking software for my testing needs.
On a similar note, do you have much experience with Analytics content experiments (not to be confused with adwords campaign experiments)? Still in beta, I think, but it's the easiest way I know of to A/B/n test macro aspects of your camps (besides 3rd party tracking maybe). Bids @ the ad group level (or KW level, if you want to get that micro), LPs, etc... You can also divert a pre-defined percentage of traffic for your tests. I think Google sets a max of 10 variations per experiment. May be worth checking out when setting up camps from scratch to get a good hold of what angles, creative variations or bid levels convert best for your offer.
Dislike ads? Remove them and support the forum:
Subscribe to Fastlane Insiders.