SES Chicago 2010: Andrew Goodman's slide deck from the ADVANCED PAID SEARCH TACTICS session. This session covered campaign expansion techniques, advanced ad testing, advanced auction theory, the proper use of relevant analytics reports, ideas for bid rules and campaign automation including automating the detection of PPC scam campaigns that can divert traffic away from the legitimate brand site and increase campaign costs.
4. Why Smart People Can Have
Stupid AdWords Accounts
“In effect, all animals are under
stringent selection pressure to be as
stupid as they can get away with.”
Richerson & Boyd, Not by Genes Alone,
2005
Dysrationalia – term coined by Keith
Stanovich, What Intelligence Tests Miss
5. The “wisdom of crowds” is just a popular way
of expressing a pretty old idea in organization
science: RATIONAL OUTCOMES don’t
immediately flow from the aptitudes of one
or two seemingly bright folks
10. The tight leash problem
What if you could program your Analytics software to look for
high CPA’s on every conceivable segment, & scale back bids?
The bias would be against volume
You’d punish randomly bad behavior
Wouldn’t be patient enough to benefit from random goodness
12. Quality Score Promotes Relevance
Ad Position determined by AdRank
AdRank = QS X Max Bid, by keyword
Quality Scores recomputed each query,
each auction
Makes money for Google
Makes search users happy
New accounts need to establish history
13. Dead Simple Takeaways (QS)
Many thorny problems with low Quality Score result
from adding keywords that are “just off” the
searcher’s true intent
Keyword tools can be misused like any other tool
A real pro shapes and moulds keyword selection and
account structure to maximize QS
Once the account is mature and generating healthy
ROI, feel free to broaden out, seeking volume &
trading off (some) relevancy
At this point Google’s “confidence level” in account is high
19. Ad testing: finding the “double
winner”
What is the “double winner”? The rare ad that
pulls higher CTR’s without hurting ROI
In landing page testing, one of the
only ways to hit on the “killer
permutation” is to test many
variations using multivariate testing
The same goes for ad copy testing
Make the big wins first
Move to MV testing
Highest possible CTR’s on ads are
paramount; affects all keyword CTR’s;
and thus affects account-wide QS
20. Key steps towards better ads
Better campaign organization = tighter
targeting (info scent)
A better ad is a higher-ROI ad? Or not!?
Tests must be statistically significant
Find initial triggers through A/B/C tests
Record hypotheses & results
Second stage: refinement (multivariate
testing)
Keep focused on “double win” genetic freaks
Master the art of “why”
22. Quality Score Booster
Create a micro promotion that boosts QS
Ordinary words like “plumber reviews” get pretty good QS
Super targeted event promos like “Toronto home show”
discounted tickets get very high CTR’s
Even if only a break-even proposition, do more of these to
inexpensively “greenlight” entire account (no downside!)
--
31. Remarketing drove the highest conversion rate and the lowest CPA for
Nexus One Display Campaign
• Conversion rate for remarketing was 2.3x higher
than the account average and CPA was 2.4x
lower
• Display ads performed the best. Remarketing
display had 3.2x higher conversion rate than the
account average for display.
• Due to high performance, the team roughly tripled
the remarketing budget in 3 months
Remarketing Case Study
32. Keepin’ em honest: PZ’s own study
Client in retail food products
Initial test: cautious; brand queries; 5 month test
Result:
Conversion rate about half the conversion rate in that
search ad group, which is *very high*
Conversion rate much higher than avg. for display
Second test on more granular ad groups
Conversion rate: 9%. Much higher than typical for display!
CPA 80% lower than allowables for those groups!
Conclusion: “nagging effect” works for cart abandoners who
still have medium-high intent – remarketing rocks!!