(F4) Measure Works - Presentation Strangeloop V5 Final
Transcript of (F4) Measure Works - Presentation Strangeloop V5 Final
How web speed
affects conversion
April 20th, Emerce Conversion
Internet is getting complex……
Everything’s tied together
The End-User: Everywhere!
Technology: Enterprise/Virtualization/Cloud
Customer Experience differs per browsers
We’re getting mobile
The end-user is in control
The eternal question: How to increase conversion?
Back to the fundamentals:
Quality of Service
Speed matters!
16 Month re-engineering Page load from 6 seconds to 1.2 Uptime from 99.65% to 99.97% 10% of previous hardware needs
Big, High-traffic site 100M impressions a day 8000 seraches a second 20-29M unique visitors a month 100M products
Shopzilla had another angle
12% increase in revenue, 25% increase in page views for
every 5 seconds of gain
As of April 9th 2010:Site speed co-determines
Google page ranking!
Speed is crucial
Average Impact of 1 second delay in response time for web users
23Source: Aberdeen Group, 2009
Let’s take a look at some Dutch websites
tested with
Page load time eTravel sites (9-15 April)
Page load time eFinance sites (9-15 April)
A closer look; Speed per browser
Time to speed up!
Shopzilla
Google Pagerank
Size and Number of Objects
Growth of average webpage size and number of objects
49.92
25.7
2.3
Average Page Size (K)Average Number of ObjectsAverage Page Size (K)Average Number of Objects
60
50
30
40
20
10
0
70
75.25
Evolving Customer Expectations
33
Response Time
8 seconds
4 seconds
2 or less
?
Global User Base
34
35
Transactional SaaS
Media Collaborative
Buy something(Amazon)
Use an app(Salesforce)
Click an ad(Google News)
Create content(Wikipedia)
Types of Sites
36
BOUNCERATE
LOSSLOSS
TIMEON
SITE
TIMEON
SITE
PAGES
PERVISIT
PAGES
PERVISIT
NUMBER
OF VISITS
NUMBER
OF VISITS
SEARCHES
TWEETS
MENTIONS
ADS SEEN
SEARCHES
TWEETS
MENTIONS
ADS SEEN
Key Metrics
AttentionAttention Engagement ConversionConversion
A/B Performance testing
10 sec
1 sec
100 ms
10 ms
Objective: Influence visitor engagement
Vis
ito
r en
gag
emen
t
-
+
39
Strangeloop AS1000
+ Tried and true optimization techniques
+ Leverage software /platform features
- Hand tune components, pages, code for performance
- No knowledge of user- Demands rare and costly
development resources
+ Offload tasks to network+ Predictable,
measurable ROI- No knowledge of
application
Strangeloop AS1000IT / Network approach
Development approach
Strangeloop AS1000
Optimize applications automatically, in real time, with a network device
A/B Performance Test Setup
Google AnalyticsGoogle Analytics
Optimize?
Decide whether
to optimize
Insertsegmentmarker
Normalcontent
Accelerated
Unaccelerated
Receivepage
Processscripts
Sendanalytics
Visitor Strangeloop Website
Technical Problem Outlined - Before
41
HTML (BackEnd)
63 Round Trips (Front End)
1.975sec 9.537sec
769KB
Strangeloop Solution - After
42
Before
Round trips: 63 Start Render: 1.975 Sec Load Time: 9.537 Sec Payload : 769 KB
After
Round trips: 8 Start Render: 1.198 Sec Load Time: 3.69 Sec Payload: 456 KB
So here’s what we discovered…
9000
0
2250
4500
6750
Traffic Levels
45
Bounce rate
46
% New Visitors
47
Average Time on Site
48
Pages per Visit
49
Conversion rate and order value
50
Currentdaily orders
Increasedconversions
Increasedorder value
Cost of performanceenhancement
* +
=ROI
(days)
( )
Justifying the investment in performance
51
$10,000 0.1607 0.0551
$50,000
* +( ) $2,158
$50,000=
23.17days=
Justifying the investment in performance