- Phil Davis (Twitter, LinkedIn)
- Erik Severinghaus (Twitter, LinkedIn)
- John Kavaliauskas (LinkedIn)
Senior Manager, Email & Data-Driven Marketing, TBC Corp
- Amanda Stewart (LinkedIn)
Senior Manager, Marketing, Entertainment Benefits Group
This panel was presented by RapLeaf (a poorly named data aggregator) and SimpleRelevance.
RapLeaf's (www.rapleaf.com) selling point is real-time data across a number of demographic points for 80% of U.S. email addresses. Load up a file (or use their API) and get back additional data points for the email address and become a more informed marketer.
SimpleRelevance (www.simplerelevance.com) makes e-commerce relevancy predictions by using publicly available data for an email address combined with the exhibited behaviors of email addresses that show similar traits, again, all with the information people share publicly online.
We have a responsibility to uphold the brand promise we made when they opted-in.
"Groupon was fun at first, but now it's all mani-pedi- it's not fun because it's not relevant." Or, to put it another way… You see this happen at a Superbowl Party: Guy goes up to Phil's wife and asks "Who are you hoping will win?" She says "I don't know who's playing and I don't really care all that much, I'm just here to enjoy the party." Guy starts talking teams and stats and after a few minutes, she's glancing over at Phil imploring him to help her opt-out of the conversation. Don't be like Guy.
Big Data or Big Headache?
Segment = assumption. Personalization is better. Don't confuse noise and signals.
Big data = non-structured unfiltered data like a twitter comment "I just test drove a Volvo" and translating that into "in market for car / values safety / has money / might be female"
Is that good enough? Can you use that to provide relevancy?
Forget "Big Data" and think "Small Data"
First-Party Data (you have this stuff already):
- Email Activity Data
- Purchase Info
- Website Activity
- Social Data (Facebook, Twitter, etc.)
Third-Party Data (you can get this stuff):
- Behavioral (bigger than your site)
The common complaint is that "We have the data, but it's all stored separately." Where are you?
- We're unable to link data at customer level.
- We're able to link, but we're unable to take action.
- We're able to take action, but it's hard to measure attribution.
Your small data has great trends and information you can use to prove (and fund) data work.
Then go to medium data by overlaying demographics over purchase info to get context. (Phil bought "My Little Mermaid" DVD for a niece. Best Buy assumed he had kids or liked Disney movies because it started promoting all kinds of Disney movies to him.) Use context to add dynamic elements to make it even more personal/special.
A list is not a thing, it's a lot of people whose interests change.
Recommended Data Points for All Marketing: Age, Gender, Location, Marital Status, Homeowner Status, Presence of Children, Income
Personalization - acquisition source is an important source of relevancy. A shoe company added recommendation block to email A/B - 50% lift with type of shoe relevant to acquisition (gym vs yoga).
First Impression: best chance to engage. But it's when we know the least. Frictionless signup - email + zip code. Careful application of data extension/append helps you learn more so that you can send more relevant emails.
Drive towards monthly revenue from email program, not from each send.
How much does it cost to get an email address? How much more to append demographics? How much more do lose from requiring demographics at signup? Or how much more to do you gain by being relevant?
Evaluate shelf-life of behavioral data. As soon as I buy a car, I'm no longer in the market for a car. But are there other opportunities in the next few months, like an oil change? Or six months, like a better insurance rate? Or detailing at 12-months?
From - are you sending from a faceless company or from a person? They'll probably stay longer if it's a person - unless it's obviously a faceless corporation impersonating a person.
Why Testing Isn't Enough
A/B testing is better than not, but it's not enough.
A/B testing is like crowdsourcing - don't miss a vital minority. Go deeper.
Even A/B testing can get stale. Just because people like red buttons today doesn't mean they will in a year.
We understand the need to test but it's hard finding the time to do it.
One company identified top selling by region, started tailoring products by region. Results: 25% increase in conversions, 75% increase in open-rates. Gained valuable insights and proved the need for more data work for relevance. (This was a quick and inexpensive win using information they already had!)
Bloomspot made some simple adjustments for relevancy and saw a 10,000x performance (not conversion) improvement from a really poor original campaign.
The Re-elect President Obama campaign used income info to determine how much they could ask someone to donate -- each person received a different email based on what they thought they could get out of the reader.
Don't be creepy
If you can make it feel relevant without feeling like you're stalking them, they will feel cared for and want to do business with you.
You can use data append clues in order to tailor without sending them an email with a Google map showing their house.
Slides are available from emailevolution.org, however, I rearranged the content from my notes for the sake of the blog post.