Test to Learn
Kristin Bond (Girl Scouts of the USA) &
Amanda Soehnlen (Highmark Health)
@asoehnlen@emailsnarketing
ALL
Tests
are
risks.
@asoehnlen@emailsnarketing
GOOD RISKS:
● Inform about audience
● Establish best practices
● Increase credibility
● Neutral to positive impact
BAD RISKS:
● Impact list health
● Hurt brand
● Break laws
● Lose lots of money
@asoehnlen@emailsnarketing
How to Come up
with what to test.
@asoehnlen@emailsnarketing
What do you want to know about your Subscribers?
● What is your goal? (Hint: it’s probably not opens or clicks - it’s probably
conversions)
● You should be optimizing for the cta.
Measure “winners” by:
Which version made the subscriber do the thing?
@asoehnlen@emailsnarketing
Your test May have inconclusive results. That’s okay.
@asoehnlen@emailsnarketing
Test Concepts
Rather Than
Content.
@asoehnlen@emailsnarketing
Test Concepts Rather than Content!
Instead of…
“I’m going to write two
subject lines I like and see
which performs better.
OMG Look at me, I’m A/B
testing like a REAL
marketer!”
Try…
“I’m going test the
concept of short vs long
subject lines and see
which performs better.
OMG Look at me, I’m A/B
testing like a REAL
marketer!”
@asoehnlen@emailsnarketing
“I’m going to test subject lines”
isn’t enough.
@asoehnlen@emailsnarketing
By testing concepts, you give
yourself opportunity to apply
results to future emails.
@asoehnlen@emailsnarketing
Ideas for Concepts to Test
● Funny vs. Serious
● Short vs. Long
● Images vs. Text
● Direct vs. Indirect
● Dollar Discount vs. Percentage
● Urgency vs. Discount
● Company’s name vs. a person’s
● Weekends vs. Weekdays
● Free shipping vs. Discount
● Text links vs. Buttons
● Early in the day vs Late
● Afternoon vs Evening
● Emojis vs text
● Gifs vs Jpegs
...Or whatever else you want to learn about your subscribers
@asoehnlen@emailsnarketing
I told you so. Or, Wow, you were right. Huh.
“If you’re right, yay, you win, and if
you’re wrong - you learn something.”
-Amanda Soehnlen
@asoehnlen@emailsnarketing
You ran a test.
You have a
winner.
What now?
@asoehnlen@emailsnarketing
What
About
The
“Losers”?
@asoehnlen@emailsnarketing
Rule of thumb:
If your results are worse than your control, use your
control (and record the test details)
(…. Your test has a control, right?)
@asoehnlen@emailsnarketing
If you’re running
a binary test…
(Do emojis work for your audience?)
@asoehnlen@emailsnarketing
Do emojis work for your audience?
@asoehnlen@emailsnarketing
Find your result, use it for everyone:
I love emojis! Oh. This sucks.
I love emojis! Yay! Text is great!
Find your result, segment the users:
60% 40%
100%
If you’re running A
OPEN-ENDED TEST…
(What’s the best send time for my users?)
@asoehnlen@emailsnarketing
When should I send my emails?
@asoehnlen@emailsnarketing
Find your result, use it for everyone:
I like emails at 7pm! Oh. This sucks.
I like emails at 7PM! 3pm!
Find your result, Continue testing variants:
2am!
55% 45%
80% 20%
If you’re testing a
concept, you have to
test more than
once.
@asoehnlen@emailsnarketing
Pop Quiz:
If this email had been a
subject line test, no
matter how it performed,
the results would be
inconclusive.
Any guesses why?
@asoehnlen@emailsnarketing
Pop Quiz:
Any guesses why?
Hint: Check the date.
@asoehnlen@emailsnarketing
The Reveal:
@asoehnlen@emailsnarketing
You may think you have a winner.
But you could be wrong.
Test the concept again. A lot.
@asoehnlen@emailsnarketing
Let’s talk about:
BLOG POSTS THAT
FEATURE
“THE BEST ”
@asoehnlen@emailsnarketing
IF IT’S NOT AN INDUSTRY BEST PRACTICE, THAT BLOG POST
DOESN’T KNOW “WHAT’S BEST.”
EVERY BLOG THAT DOESN’T USE THE CAVEAT “FOR MY
AUDIENCE…”
QUESTION. THE. RESULTS.
IT MAY NOT BE APPLICABLE TO YOU OR YOUR AUDIENCE.
@asoehnlen@emailsnarketing
Everything you
know could be
WRONG.
@asoehnlen@emailsnarketing
That’s why you
test.
@asoehnlen@emailsnarketing

Test to Learn

  • 1.
    Test to Learn KristinBond (Girl Scouts of the USA) & Amanda Soehnlen (Highmark Health) @asoehnlen@emailsnarketing
  • 2.
  • 3.
    GOOD RISKS: ● Informabout audience ● Establish best practices ● Increase credibility ● Neutral to positive impact BAD RISKS: ● Impact list health ● Hurt brand ● Break laws ● Lose lots of money @asoehnlen@emailsnarketing
  • 4.
    How to Comeup with what to test. @asoehnlen@emailsnarketing
  • 5.
    What do youwant to know about your Subscribers? ● What is your goal? (Hint: it’s probably not opens or clicks - it’s probably conversions) ● You should be optimizing for the cta. Measure “winners” by: Which version made the subscriber do the thing? @asoehnlen@emailsnarketing
  • 6.
    Your test Mayhave inconclusive results. That’s okay. @asoehnlen@emailsnarketing
  • 7.
  • 8.
    Test Concepts Ratherthan Content! Instead of… “I’m going to write two subject lines I like and see which performs better. OMG Look at me, I’m A/B testing like a REAL marketer!” Try… “I’m going test the concept of short vs long subject lines and see which performs better. OMG Look at me, I’m A/B testing like a REAL marketer!” @asoehnlen@emailsnarketing
  • 9.
    “I’m going totest subject lines” isn’t enough. @asoehnlen@emailsnarketing
  • 10.
    By testing concepts,you give yourself opportunity to apply results to future emails. @asoehnlen@emailsnarketing
  • 11.
    Ideas for Conceptsto Test ● Funny vs. Serious ● Short vs. Long ● Images vs. Text ● Direct vs. Indirect ● Dollar Discount vs. Percentage ● Urgency vs. Discount ● Company’s name vs. a person’s ● Weekends vs. Weekdays ● Free shipping vs. Discount ● Text links vs. Buttons ● Early in the day vs Late ● Afternoon vs Evening ● Emojis vs text ● Gifs vs Jpegs ...Or whatever else you want to learn about your subscribers @asoehnlen@emailsnarketing
  • 12.
    I told youso. Or, Wow, you were right. Huh. “If you’re right, yay, you win, and if you’re wrong - you learn something.” -Amanda Soehnlen @asoehnlen@emailsnarketing
  • 13.
    You ran atest. You have a winner. What now? @asoehnlen@emailsnarketing
  • 14.
  • 15.
    Rule of thumb: Ifyour results are worse than your control, use your control (and record the test details) (…. Your test has a control, right?) @asoehnlen@emailsnarketing
  • 16.
    If you’re running abinary test… (Do emojis work for your audience?) @asoehnlen@emailsnarketing
  • 17.
    Do emojis workfor your audience? @asoehnlen@emailsnarketing Find your result, use it for everyone: I love emojis! Oh. This sucks. I love emojis! Yay! Text is great! Find your result, segment the users: 60% 40% 100%
  • 18.
    If you’re runningA OPEN-ENDED TEST… (What’s the best send time for my users?) @asoehnlen@emailsnarketing
  • 19.
    When should Isend my emails? @asoehnlen@emailsnarketing Find your result, use it for everyone: I like emails at 7pm! Oh. This sucks. I like emails at 7PM! 3pm! Find your result, Continue testing variants: 2am! 55% 45% 80% 20%
  • 20.
    If you’re testinga concept, you have to test more than once. @asoehnlen@emailsnarketing
  • 21.
    Pop Quiz: If thisemail had been a subject line test, no matter how it performed, the results would be inconclusive. Any guesses why? @asoehnlen@emailsnarketing
  • 22.
    Pop Quiz: Any guesseswhy? Hint: Check the date. @asoehnlen@emailsnarketing
  • 23.
  • 24.
    You may thinkyou have a winner. But you could be wrong. Test the concept again. A lot. @asoehnlen@emailsnarketing
  • 25.
    Let’s talk about: BLOGPOSTS THAT FEATURE “THE BEST ” @asoehnlen@emailsnarketing
  • 26.
    IF IT’S NOTAN INDUSTRY BEST PRACTICE, THAT BLOG POST DOESN’T KNOW “WHAT’S BEST.” EVERY BLOG THAT DOESN’T USE THE CAVEAT “FOR MY AUDIENCE…” QUESTION. THE. RESULTS. IT MAY NOT BE APPLICABLE TO YOU OR YOUR AUDIENCE. @asoehnlen@emailsnarketing
  • 27.
    Everything you know couldbe WRONG. @asoehnlen@emailsnarketing
  • 28.

Editor's Notes

  • #22 First off - are you sure you even have a winner? This email probably had an absolutely bonkers open rate. But if it had been part of a SL test, it would have been inconclusive. Subject line was “Dress in Color.” Any guesses why?
  • #23 First off - are you sure you even have a winner? This email probably had an absolutely bonkers open rate. But if it had been part of a SL test, it would have been inconclusive. Subject line was “Dress in Color.” Any guesses why?
  • #24 First off - are you sure you even have a winner? This email probably had an absolutely bonkers open rate. But if it had been part of a SL test, it would have been inconclusive. Subject line was “Dress in Color.” Any guesses why?