Post on 23-Jan-2018
transcript
#SMX #11B @AndreasReiffen
About…
• Data-driven online advertising strategist
• Online retail expert
• Entrepreneur
• Over €3 billion in customer revenues
last year
• SaaS product for Google Shopping &
Search
• 130 true experts in their field
• Offices in Germany & UK, new office in
NYC
… me … crealytics & camato
#SMX #11B @AndreasReiffen
Topics we will cover today..
Economies of
Google shopping Role of price Image testing
Product title
testing
How does price
influence
performance?
How big a role
does price play in
product ranking?
CPCs have
decreased!
What will happen
next?
How can you test
images properly?
What is the effect
of image
changes?
Can you
manipulate titles
and get more
relevant user
queries?
#SMX #11B @AndreasReiffen
Global Shopping growth 2016 Shopping Click Share 2014 - 2016
In 2016 Shopping clicks kept increasing while CPCs decreased –why?
Clicks CPC
-15%
36%
*Alphabet earnings report
74%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Q2Q1 2016
Q2 Q4 Q4Q4Q3 Q3 Q3Q2Q1 2015
Q1 2014
Shooping
Text Ads
*Internal data US + EU
#SMX #11B @AndreasReiffen
Click Share by device YoY
Increasing mobile click share
As predicted by Google, mobile click sharein Shopping has increased massively overthe past few years, so that since 2016 themajority of clicks is generated by mobile
rather than desktop
100
84
16
49
51
30
2016
100
70
100
62
38
100
2015
Desktop
20172014
Mobile
#SMX #11B @AndreasReiffen
Ad space pushing organic space Resulting in stronger growth
More advertising space
Paid
+78%
+16%
Organic
#SMX #11B @AndreasReiffen
Impression volume for generic terms (M)
More Shopping results for generic queries
In 2016 Google startedincreasing the amount ofShopping Impressions forgeneric terms, leading toan overhead in available
queries.
Q4 2015
Q3 2016
66
Q4 2016
Q3 2015
Q1 2015
Q2 2016
8
Q1 2016
10
9
6
7
4
Q2 2015
2
6
2
Q1 2015
5
Q4 2015
Q3 2015
3
Q2 2016
5
Q2 2015
2
Q1 2016
Q4 2016
Q3 2016
5Category termsDesigner terms
#SMX #11B @AndreasReiffen
Paid sessions per user YoY
Navigation happening on Google rather than website
46% shopping conversions have 3+ shopping clicks
29% text ads conversions have 3+ text adsclicks
GoogleWebsite
Purchase1.39
1.48
1.73
2014
+17%
2016
1.41
Text Ads
Shopping
#SMX #11B @AndreasReiffen
Shopping vs Text Ads CPC & RPC CR by device
Shopping more profitable despite higher overall CPCs
100100
162
121
+21%
+62%
RPCCPC
Text Ads
Shopping
97%
Text Ads Mobile
61%
189%
Shopping Mobile
Text Ads Desktop
Shopping Desktop
100%
#SMX #11B @AndreasReiffen
Your key take aways
Strong ongoing growth
of ad inventory supply is
leading to CPC deflation
Be faster than your
competition!
Move budget to
Shopping now!
CPCs still higher than
on Text Ads but due to
high CR Shopping
more efficient than
Text ads
#SMX #11B @AndreasReiffen
Impressions and clicks in Google Shopping are often very sensitive to price changes
Clicks drop off after product price increase Chart Info
5% price increase coincides with a 60% decrease in clicks
0
20
40
60
80
100
120
0
10
20
30
40
50
60
Pri
ce in
£
+5%
clicksown price Google product category: Apparel & Accessories > Shoes > Sneakers
Brand:
Days
Clicks
Price
#SMX #11B @AndreasReiffen
Cheap products generate the lion‘s share of total traffic
Traffic of similar products within one shop Key insights
Cheap products generatedmuch more traffic despitesimilar number of productsin both groups
Imps
134%
4.282.611
1.828.412
103.803
135%
Clicks
44.122
2.047
9%
# of products
1.876
expensive products cheap products
#SMX #11B @AndreasReiffen
Despite getting more traffic, the performance of the cheaperproducts is much more efficient.
Performance of similar products within one shop Key insights
Number of orders throughcheap products almost 3 times higher.
Higher CR resulting in CPO being ~30% lower
40
28
-29%
CPO Conversions
1,042
+280%
274
61%
0.6%
CR
1.0%
expensive products cheap products
#SMX #11B @AndreasReiffen
Price is very important in terms of ranking but there areother factors as well
$Retailer Rating
New retailers will be rateddown, seller ratings help
Performance History
More history leads to more impressions
Price
Users favor cheaperproducts, so these achievebetter positions on google
Bid
Bid is important but can be outweighed by product
price
According to Google, price is not an explicit factor in their algorithm,but maybe taken into account through machine learning if users favor lower prices
#SMX #11B @AndreasReiffen
Position of cheapest offer
Product price has a massive impact on position!
Pos 1
Pos 2 14%
65%
Cheapest
Pos 5 +
Pos 4 4%
7%Pos 3
11%
* Data from price comparison provider based on > 3000 queries
Is Google shopping a price race to the bottom?
In 65% of all cases in our sample, theproduct with the most competitiveprice was ranked on position 1, no
matter the retailer!
#SMX #11B @AndreasReiffen
The way out: We analysed what people actually boughtwhen they clicked on a product ad
Same Designer Different Designer
purchaseclick
34%
16%
14%
15%
21%
36%
Dif
fere
nt
Ca
teg
ory
Sa
me
Ca
teg
ory
#SMX #11B @AndreasReiffen
Your key takeaways
Don’t overbid on
expensive
products, rather
consider price
changes
Price and bid
management will
be merged one
day
Discounting only a
few select
products could be
a killer strategy
$.47$.47-50%
#SMX #11B @AndreasReiffen
Challenges to image testing
Do I have cassified
images? Time lag
SERPs vs shopping
tab
A B Shopping tab:
Products clustered intoone offer list.
Not necessarilyyour image!
Time to index new image:
• Up to 72 hours when URLs are changed
• Up to 6 weeks if URLs remains the same (server-side image swap)
Know when you measure “A”. More data required. Long testing cyle.
Change images consistently from A to B across entire category
#SMX #11B @AndreasReiffen
About the data
1
2
3
2 fashion retailers: 1800 images changed in total
> 40m impressions
Products were ignored if not available at all times
target metrics: CTR (user), impressions (Google)
#SMX #11B @AndreasReiffen
Change images in “test“, use “control“ as baseline. Compare KPIs throughout test to eliminate seasonality.
Test
Control
Before During After
#SMX #11B @AndreasReiffen
Model image 2.6% stronger CTR Model image 27% stronger CTR
Two tests, only one has a clear positive result
100.0%127.0%100.0% 102.6%
Images
CTR
Images
CTR *
Not significant Significant
* CTR test vs control, with original image set to 100%. First test > 5m imps, second test >40m imps.
#SMX #11B @AndreasReiffen
Image test seems to work when image sets you apart
Images
Competition
Images
Competition
Image type varies no effect Stand out positive effect
BA BA
#SMX #11B @AndreasReiffen
Your key take aways
Image testing can havean effect:
Stand out when many offer the same product
There are areas where testing can have a more significant
impact and the return on effort is much bigger.
Images are important for Google to match
queries
mountain
#SMX #11B @AndreasReiffen
In feed optimisation, titles are most promising.
Product titles Google CategoryProduct description
Evening dressby TFNC…
before destroyed
100% 98%
Baseball capby Nike…
Shoes > …> Athletic Shoes
118%100%
destroyedbefore
Clothing > …> Rain Suits
Lipsy Lace DressNavy blue..
Party DressesNavy Blue Lipsy…
title changed
original
100%
960%
Imps Imps Imps
#SMX #11B @AndreasReiffen
Testing further strategies to determine impact of title changes
1
2
3
4 accounts from 3 clients & 3 verticals
16.8k titles changed
5.2m Impressions generated in total
#SMX #11B @AndreasReiffen
What we found out:
Unrelated Terms
Terms with volume on google but no relation to product - therefore don’t
show up in product titles.
Seasonal terms: Christmas/ Black Friday General terms: gift/ fashion/ online
Google is too smart: Unrelated terms have no impact
Existing Queries
High share of voice existing queries, not included in titles.
Synonyms: Laptop Notebook
Trainers Sneakers
Slight impact but not worth the effort.
Under-Represented
Queries1. Low share of voice existing, not
included in titles.
2. High traffic Text Ads terms, not receiving Shopping Impressions
Success: Significant increase in volume
#SMX #11B @AndreasReiffen
Under-represented queries – Strong positive impact
100100
111
134
+34%
+11%
Account LevelQuery Level
After
Before
Impressions
This strategy has proven to be the mostsuccessful.
As the high text ads traffic on these termsalready indicated, the impression volume
on Shopping had potential for growth.
#SMX #11B @AndreasReiffen
Low/mid-level traffic queries with market potential are thebest choice
Title change:
+Designer Trainer
Share of available
volumeImpressions on query
Reebok Z Pump Fusion Training - Men
Designer Trainer by Reebok Z Pump Fusion Training
across 39 products KW planner Traffic / m before
Traffic / m after
Example
100100 116
7.100
Query Level
+16%
+7.000%
Account Level
After
Before
#SMX #11B @AndreasReiffen
Including existing queries can be succesful
Jordan 1 Flight 4 Premium Space Jam
Jordan Kids - Jordan 1 Flight 4 Premium Space Jam
across 26 products
Example
100100120
167
Query Level
+20%
+67%
Account Level
After
Before
KW planner Traffic / m before
Traffic / m after
Title change:
+Jordan Kids
Share of available
volumeImpressions on query
#SMX #11B @AndreasReiffen
Your key take aways
Title optimisation works.
Up to 30% more Impressions!
Don’t overwrite important terms!
(it will eat up your gains)
Focus onproduct-related
andlow share of voice
#SMX #11B @AndreasReiffen
For a detailed step by step guide on how to define which terms to add to your titles,
visit our blog at:
www.crealytics.com/blog/title-changes-how-to/