<< by on April 26th, 2012
If you missed the Advanced Track this morning at SMX Toronto, here are the highlights from the Paid Search: Advanced Testing and Reporting Panel. Brad Geddes, Founder of Certified Knowledge, and Jeff Ferguson, CEO of Fang Digital Media, lead this panel. This session centered around the actual process of designing, implementing and measuring the results of paid search tests.
Brad Geddes kicked off the panel discussing the best ways to evaluate testing and which metrics create a conversion
- Clicks – CPC
- Impressions & clicks – CTR
- Average sale amount
- Conversion metrics -Ignores the impressions and clicks
- CPA – Ignores the impressions and clicks
- ROI/ROAS – Percentage, ignores volume
- Conversion per impression -Doesn’t take in average sale amounts
Metrics should start at the impression and accommodate CPCs & cost. Take in account conversions and sale amount – sale amount can vary by ad type.
The key metric to focus on: Revenue/Profit per Impression
PPI = (revenue – cost)/Impressions
- If you have margins/hard costs, etc. remove those before calculating
Extrapolate to see profit. Multi-sales per customer? – don’t forget lifetime value
Testing Ad copy – start with people
- What do people want?
- What do people avoid?
People don’t see bidding, time of day bidding, or keyword targeting – people only see the ad copy. Focus on the ad.
Test pricing – useful for testing in industries that are – price conscious, highly competitive, large price ranges, shopping comparison.
- Changing prices in ad copy, cause ad copy to be reviewed and delay ads running.
- Test a percentage off to prevent frequent ad copy changes
Discounts – bigger is better, people often focus on the biggest and smallest numbers
Calls to Action – make a call to action a benefits statement, ‘Sign up for our newsletter’ OR ‘Signup for Powerful Marketing Tips.’
- Powerful words – ‘secret’, helps CTR but not always CR and revenue.
Information – people in the research process: exclusive info that others can’t share. This can be powerful
Local Information – are you marketing to locals or non-locals?
- Locals more familiar with specific neighborhoods and is more targeted.
- Refer to local land marks, locals are familiar, tourists probably aren’t.
- Non-locals need more general info.
Questions – what question did the searcher want answered?
Testimonials – wisdom of the crowds. Can’t use, ‘best’, ‘#1’, etc. in ad copy unless it is verified by a 3rd party. This can help you stand out from competition, they can only use general terms like ‘top’.
Negative information – There’s a reason news headlines are often negative.
Even more items to test –
- Shipping method – especially for non US customers
- Customer benefits
- Unique Value Props
- Product/Service Features
- Symbols – Copyright, trademark
- One line vs. 2 link ad copy
- Be creative!
Testing is easy – opt for Google to rotate ads – 3 ways to rotate ads: optimize for clicks, optimize for conversions, or rotate evenly.
Wait for results – evaluate and choose the winner. Any data anomalies? Just a little date analysis makes a big difference in accounts.
Jeff Ferguson – Widen your gaze. With so much data, it’s easy to focus on just one number.
Digging into the data – day of week/ week of month testing. Look beyond the simple day of week
- Early in the month, much higher performance, but the rare 5th week of the month often does poorly
- Look at individual campaigns by week of month. Performance wasn’t universal
- Dig into keyword/ad group level data
- Dig into intra month trends, summers, winters, seasons, intra year seasonality
- Remove any big spikes from the data to prevent skewing the numbers
- Be proactive with data, but don’t get sucked into small adjustments daily, move twards analyzing the data to get ahead of trends
- Robots are okay but don’t become a slave to their charms
- Look for local trends within the data –
- Seasonality – buying seasons, spring thaw, paydays
- Upcoming events – festivals, concert schedule
- Find your own Week of Month
Geo Silos – for more than just local targeting
Dig into individual metro area data – analyze major DMAs
- Comparing DMAs where company has brick and mortar stores vs. no local presence
- The big strategy – optimize based on geo, despite not being a ‘local’ company