Date post: | 08-May-2015 |
Category: |
Business |
Upload: | marketingexperiments |
View: | 2,493 times |
Download: | 0 times |
How Many Columns Should I Use? How using the best page layout led to a 681% relative increase
We’re sharing on Twitter!#WebClinic
Today’s speakers
@jonpowell31
Jon PowellSenior Manager Research and StrategyMECLABS
Marketing research practitioner with 7 years of hands-on experience gained through managing hundreds of A/B and multivariate tests for e-commerce, health care, financial and education Research Partners.
Today’s topic
Given a choice between these particular page templates, which would you choose for your offer pages? Why??
A B C D
What is the highest performing layout (number of columns) for my webpages? How do I know which one to choose?
?
Today’s topic
Background: A large technology company selling software to small businesses.
Goal: To significantly increase the number of software purchases from paid search traffic (branded terms).
Research Question: Which column layout will generate the highest rate of software purchases?
Test Design: A/B multifactor split test
Experiment ID: TP1736aRecord Location: MECLABS Research LibraryResearch Partner: [Protected]
Experiment 1: Background
COLUMN #1 COLUMN #2
Experiment 1: Control2-Column Layout
COLUMN #1
Experiment 1: Treatment1-Column Layout
COLUMN #1 COLUMN #2COLUMN #1
2-Column Layout 1-Column Layout
Experiment 1: Side-by-side
Design Order Rate Revenue/Visit
Control – 2 Columns 0.20%* $0.50*
Treatment – 1 Column 1.58%* $3.56*
% Relative Change: 680.6% 606.7%
Relative Increase in Orders681%The one-column treatment design increased branded search orders by 680.6% and revenue per visit by 606.7% when tested against the two-column control design.
Experiment 1: Results
*Absolute performance rates and numbers have been anonymized
Why did the single-column page layout have a significant increase in results??
Audience question2-Column Layout 1-Column Layout
Experiment: Results
“The eye path of the page is confusing to users … ”
“ … by providing a linear eye path, distractions will be removed and will cause an increase in conversion.”
From the Test Protocol:
*Certain sections of content have been anonymized for the Partner
266%In Clicks
Protocol IDs: TP1306Similar findings
2-Column Layout 1-Column Layout
Example: Main offer page
22%In Completed Orders
Protocol IDs: TP1356, TP1407Similar findings
2-Column Layout 1-Column Layout
Example: Transaction page
Is there an underlying theory for choosing the highest performing page layout? ?
Background: A large technology company selling software to small businesses.
Goal: To significantly increase the number of software purchases from paid search traffic (non-branded terms).
Research Question: Which column layout will generate the highest rate of software purchases?
Test Design: A/B multifactor split test
Experiment ID: TP1736bRecord Location: MECLABS Research LibraryResearch Partner: [Protected]
Experiment 2: Background
COLUMN #1 COLUMN #2COLUMN #1
2-Column Layout 1-Column Layout
Experiment 2: Side-by-side
Design Order Rate Revenue/Visit
Control – 2 Columns 5.69%* $19.78*
Treatment – 1 Column 0.86%* $6.53*
% Relative Change: -84.9% -67.0%
Relative Decrease in Orders85%The one-column treatment design decreased non-branded search orders by 84.9% and revenue per visit by 67% when tested against the two-column control design.
Experiment 2: Results
*Absolute performance rates and numbers have been anonymized
Why did the winning page layout suddenly underperform in the same channel??
Experiment 2: Results
To help answer this question, we accessed our library of case studies for a deeper study and isolated tests where page layout was a primary variable.
We then conducted a meta-analysis of those Test Protocols to uncover patterns in performance.
Plus many more Protocols …
TP1457TP1460 TP1621 TP1746
TP1306 TP1356 TP1560 TP1633
TP1267 TP1315 TP1639 TP1268
TP1499 TP1658 TP1636 TP1203
TP1635 TP1498 TP1165 TP1475
TP1192 TP1182 TP1469 TP1091
TP1460 TP1585 TP1214 TP1503
TP1422 TP1500 TP1488 TP1616
TP1511
Initial findings
There appeared to be performance patterns for each page in the conversion path …!
Initial findings
However, the patterns were not clear due to numerous outliers.!
3 Columns 2 Columns
Protocol IDs: TP1268OutlierExample: Transaction page
54%In Revenue per Order
15%In Revenue per Order
Homepage traffic Shopping cart traffic
60%In Conversions
Protocol IDs: TP1457Outlier
2-Column Layout 1-Column Layout
Example: Main offer page
3 Columns 1 Column
7.5%In Searches
Holiday Traffic
5.8%In Searches
Everyday Traffic
Protocol IDs: TP1165, TP1182OutlierExample: Directory page
Is there an underlying theory for choosing the highest performing page layout? ?
Back to the original question
ChannelMain Offer
Page
Checkout Path
Category Page
Thank You Page
Homepage
Directory Pages
When we took a closer look at TP1737a-b, each winning page is essentially in the same position in the conversion funnel.
Experiment: Analysis
They are NOT, however, in the same position in the customer funnel.
Experiment: Analysis
(Mi)YES
(Ma)YES
Y
Val
ue
Pro
po
siti
on
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
Non-Branded Winner
Branded Winner
A Web search using brand language indicates prior knowledge of the value proposition.
A generic Web search does not, requiring more clarity of value.
(Mi)YES
(Ma)YES
Y
Val
ue
Pro
po
siti
on
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
Meta-analysis results
We recategorized each Test Protocol in the analysis according to its stage in the customer thought sequence (as opposed to the conversion path).
In doing this, we discovered a single pattern that encompassed all Test Protocols analyzed.
(Mi)YES
(Ma)YES
Y
Val
ue
Pro
po
siti
on
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
More
Meta-analysis results
At the beginning of the thought sequence, the highest performing pages often have three or more columns that are evenly weighted.
(ex: 33%/33%/33%)
Even
Less
Number of Columns
Uneven
Column Weight
(Mi)YES
(Ma)YES
Y
Val
ue
Pro
po
siti
on
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
(Mi)YES
As the thought sequence progresses, column counts in high-performing pages drop. If there are two columns, they are unevenly weighted.
(70%/30% or 80%/20%)
More Even
Less Uneven
Column Weight
Meta-analysis resultsNumber of Columns
On the ideal number of columns
FKey Principles
1. First, we must understand there is no single answer to the question, “How many columns should I use?” Sometimes “less” is optimal; and sometimes “more” is optimal.
2. To begin answering this question, we must understand where the customer is in the mental (not physical) conversion funnel. From there, we can apply two underlying principles:
1. The more we know about what the visitor wants, the more we should focus on that desire (less columns; unequally weighted).
2. The less we know about what the visitor wants, the more we should enable the them to self-select (more columns; equally weighted)
Live OptimizationColumns
Live Optimization
Tracky
http://bit.ly/1aJj4sL #WebClinic
Primary Audience: Business and consumersPrimary Objective: Awareness
Live Optimization
Santa Fe Way
http://bit.ly/1c3zQDN #WebClinic
Primary Audience: Health enthusiastsPrimary Objective: Information page
Live Optimization
Primary Audience: B2CPrimary Objective: Lead gen/sales
Windstream
http://bit.ly/1f6PyP1 #WebClinic
Live Optimization
Primary Audience: Potential donorsPrimary Objective: Raise funds
Environmental Defense Fund
http://bit.ly/1aqws7k #WebClinic
Live Optimization
StudyAbroad.com
http://bit.ly/KGhAG1 #WebClinic
Primary Audience: StudentsPrimary Objective: Info page
Live Optimization
Declaring Bankruptcy
http://bit.ly/1biojwl #WebClinic
Primary Audience: B2C Primary Objective: Lead generation
Live Optimization
Primary Audience: MembersPrimary Objective: Fundraising
Jewish Federation
http://bit.ly/1f6PyhV #WebClinic
Live Optimization
Primary Audience: B2CPrimary Objective: Lead generation
IHG
http://bit.ly/JH3dkj #WebClinic
Live Optimization
Applied Inspirations
http://bit.ly/1dvTJ24 #WebClinic
Primary Audience: B2CPrimary Objective: Sales
Background: A large, well-known audio technology and engineering company offering professional and personal audio products.
Goal: To significantly increase the number clicks from a promotional email.
Research Question: Which email call-to-action copy will produce the greatest clickthrough rate?
Test Design: A/B single factorial split test
Experiment ID: TP2136Record Location: MECLABS Research LibraryResearch Partner: [Protected]
Next Clinic: Background
Next Clinic: Email template
Next Clinic: Email template
Next Clinic: Version A
Next Clinic: Version B
Next Clinic: Side-by-Side
Version A Version B
“Shop Now” “View Details”
Which treatment won?
Live Feb. 12 at 4 p.m. EST
• How can I get the most clickthrough for the least effort?• Which call-to-action copy works best for promotional emails?• How does copywriting change for newsletters and nurture-
based emails?
Join the live 35-minute Web clinic
To see the results
To join live, register at the link below:
MarketingExperiments.com/TwoWordsEmail
See how you can conduct research with us
MECLABS conducts rigorous experiments in the new science of optimization. We apply our discoveries to help leaders optimize the financial performance of their sales and marketing programs.
Learn more about how you may be a fit for a MECLABS Research Rartnership:
• Select Research Partnership Opportunities on the post-webinar survey
• Contact us [email protected]
x