“Could too much A/B testing be detrimental to PPC?” asks Matt Van Wagner of Find Me Faster at a recent SEMNE event. Follow Matt Van Wagner’s SEMNE slideshow to see how to obtain advertising success on the internet.
A/B testing: Which message resonates best with your target audience?
- Run ads to create performance profile
- Find out which ad is the best
Reasons why AdWords fail
- Keywords don’t match queries
- Ad impressions doesn’t equal number of searchers
- Ad rotation isn’t even/balanced
- A/B testing will give you the best ad, not designed to give you your highest performing ad group
- As many of 50% of searches are repeats, and they happen in hotspots, search up to 20 times
- 5-20 impressions could be just one person
- Quality score impacts ad rotation, so rotation might not be even
Winning ads perpetuate themselves
- A few well-performing ads will be more effective, we forget that individuals with personal preferences are doing the searches
- Humans can out-smart the algorithms
- “Save 20%” message vs “eco” message: these are two different demographics you’re speaking with
- Keeping multiple ads gives a fighting chance to engage with divergent populations. Why would you only write ads for men and cut out women?
- 5 well-performing ads will outpace one top-seller
Google is trying to make money, so if you didn’t click on the first ad, it’ll show you another version
- After 1 search you’ll see different results, by the time visitors search 4 times there’s a 93.75% chance that visitors are seeing both ads
Bottom line: Don’t be so quick to kill off ads that are performing until you have a top 5 and keep testing
- Don’t try to reduce to 1 ad
- Test more combinations
- Ad sets can mask the performance of poorly performing ad groups
- Set ad rotation to even
Broad Match Modifier
- Create keywords that give you more control than broad, more flexibility than phrase match
- Use + sign in front of broad-match keyword
- Now available in the US, think how can use broad match to direct traffic to targeted places.
[New Post] A/B testing PCC: Is too much detrimental? – via #twitoaster http://blog.designatededitor.com/2011/02…