Optimize Google Shopping with the help of Euclid

Optimize Google Shopping with the help of Euclid #ecommercial 16 @whoop_app

Introduction Chris Scharmüller Head of Strategic Customer Development & Sales Whoop! / smarter-ecommerce #ecommercial 16 @whoop_app

Euclid (of Alexandria) or “Father of Geometry” #ecommercial 16 @whoop_app

Black Box Bidding The solution: Orthogonalization 0, 55$ #ecommercial 16 One „all products“ target for your whole assortment @whoop_app

First Dimension: Items One Ad Group per SKU $0. 15 $0. 32 $0. 05 $0. 17 #ecommercial 16 $0. 57 $0. 10 $0. 11 @whoop_app

Second Dimension: Devices The ad group structure enables you to set device modifiers (= bids) for every item $0. 15 $0. 32 $0. 05 $0. 17 #ecommercial 16 $0. 57 $0. 10 $0. 11 @whoop_app

Third Dimension: Query Types (or intent) Short explanation of the underlying concept Nike Free 5. 0 Nike Men’s Running Shoes Nike Free 5. 0 … 99. 90 € shop. co. uk Running Shoes CAMPAIGN GROUP PRIORITY NEG. KEYWORDS BIDS Products low none very high Brands medium product-terms

like “free 5. 0” high Other high productterms

brand terms low #ecommercial 16 BUDGET Shared Budget @whoop_app

Characteristics of shopping queries Low Med High Query Word Count 1 -2 2 -4 3 or more Brand Reference not likely very likely most likely Intent-based Query Components not likely (top-rated. price. comparison) likely (buy. online shop. financing options) Match to Inventory very many products match only a handful of 
products match Last Click Conversion Rates low medium high Traffic Volume high very low to low medium to high #ecommercial 16 @whoop_app

Third Dimension: Query Types Based on Query Sculpting $0. 15 $0. 32 $0. 05 $0. 17 #ecommercial 16 $0. 57 $0. 10 $0. 11 @whoop_app

Thanks #ecommercial 16 @whoop_app

How to apply the concept of the purchase funnel to Google Shopping? PRODUCT Nike Free 5. 0 Running Shoes Neon CLICKS CONV CR VPC CPC 1500 19 1. 26 % € 1. 26 € 0. 58 Nike Free 5. 0 … 99. 90 € shop. co. uk Target-ROAS: 5 Actual ROAS: 2. 17 Action: Reducing the bids The problem of Google Shopping: Every type of query is treated the same! #ecommercial 16 @whoop_app

Shopping Campaigns are auto-targeted There are no keywords for Shopping Campaigns. but you are able to use negative keywords! #ecommercial 16 There are no keywords for Shopping Campaigns. but you are able to use negative keywords! @whoop_app

Query Sculpting Solving a big problem of Google Shopping SEARCH TERMS (= Queries) Nike Free 5. 0 … 99. 90 € shop. co. uk CLICKS CONV CR VPC CPC ROAS Nike Free 5. 0 100 3 3% € 3 € 0. 5 6 Buy Nike Men’s Running Shoes Free 250 12 4. 8% € 4. 8 € 0. 4 12 Men’s Running Shoes 300 0 0% € 0. 6 0 Running Shoes 600 1 0. 16% € 0. 16 € 0. 7 0. 22 Nike Men’s Running Shoes 250 3 1. 2% € 1. 2 € 0. 5 2. 4 #ecommercial 16 @whoop_app

Explicit Query Sculpting Advantages Great way to manage query behavior Flaws Patterns can be very heterogeneous and won’t be applicable on the whole assortment Also those patterns have to be researched in the first place and managed continuously Potential hidden performers will stay in the “other”-Campaign #ecommercial 16 @whoop_app

Query Sculpting SEARCH TERMS (= Queries) Nike Free 5. 0 … 99. 90 € shop. co. uk CLICKS CONV CR VPC CPC ROAS Nike Free 5. 0 100 3 3% € 3 € 0. 5 6 Buy Nike Men’s Running Shoes 250 12 4. 8% € 4. 8 € 0. 4 12 Men’s Running Shoes 300 0 0% € 0. 6 0 Running Shoes 600 1 0. 16% € 0. 16 € 0. 7 0. 22 Nike Men’s Running Shoes 250 3 1. 2% € 1. 2 € 0. 5 2. 4 Think of queries (and query components) having a different likeliness of conversion. #ecommercial 16 @whoop_app

3 Buckets for your Search Queries QUERIES Nike Free 5. 0 … 99. 90 € shop. co. uk CLICKS CONV CR VPC CPC ROAS SCORE Nike Free 5. 0 100 3 3% € 3 € 0. 5 6 high Buy Nike Men’s Running Shoes 250 12 4. 8% € 4. 8 € 0. 4 12 high Men’s running shoes 300 0 0% € 0. 6 0 low Running Shoes 600 1 0. 16% € 0. 16 € 0. 7 0. 22 low Nike Men’s Running Shoes 250 3 1. 2% € 1. 2 € 0. 5 2. 4 med We use a score for defining the likeliness of a conversion. The score is used to group the queries into 3 buckets. #ecommercial 16 @whoop_app

Funnel Buckets CLICKS CONV CR VPC CPC ROAS ACTION 
(for Target-ROAS > 5) high
(lower funnel) 350 15 4. 29% € 4. 29 € 0. 43 9. 98 Large Increase of Bids Med
(mid funnel) 250 3 1. 2% € 1. 2 € 0. 5 2. 4 Small Increase of Bids Low
(upper funnel) 900 1 0. 11% € 0. 11 € 0. 66 0. 16 Large Decrease of Bids You can see the results. #ecommercial 16 @whoop_app

Tactic Flow for Shopping Campaigns Campaign 3 ”low” Mostly exact terms with a low conversion probability Mostly phrases consisting out of 2 -6 words having a high conversion probability Campaign 2 ”high” Average Terms and pool for gathering new queries Campaign 1 ”med” Traffic #ecommercial 16 @whoop_app

Campaign Structure - The Concept CAMPAIGN PRIORITY NEGATIVE KEYWORDS Campaign 3 ”low” Low Priority none Campaign 2 ”high” Medium Priority Low Performing Queries as Negative Exact High Priority Low Performing Queries as Negative Exact Campaign 1 ”med” High Performing Queries as Negative Phrase Traffic #ecommercial 16 @whoop_app

Term Scoring Performance metrics Term length Color reference Brand reference tdf/idf weight Many others SCORE #ecommercial 16 @whoop_app

Term Scoring - a very simplified example SCORE = Word. Count + if(Brand. Ref = True. 2. 0) + Normalized. CTR/2 + Normalized. Conversion. Rate*2 + other Arguments 3. 32 = 2 0 1 Running Shoes 20. 6 = 5 2 2% CTR 4 Buy Nike Men’s Running Shoes 8% CTR 0. 32 0. 16% CR 9. 6 4. 8% CR further Reading: http: //nlp. stanford. edu/IR-book/pdf/irbookonlinereading. pdf #ecommercial 16 @whoop_app

Scored terms based on an advanced model (Potential) High performers - Last 30 Days Clicks CR Cost Score asics gt 2000 4 235 6. 38% 81. 73 154. 31 asics gt 2000 4 men 85 4. 71% 31. 98 145. 48 adidas zx 750 522 4. 79% 172. 78 142. 40 asics gel cumulus 17 211 4. 27% 97. 1 131. 10 asics gt 2000 4 women 89 3. 37% 37. 76 124. 15 asics gt 2000 276 3. 26% 127. 14 121. 10 nike metcon 2 251 3. 19% 63. 63 120. 14 asics kayano 22 132 3. 03% 41. 93 120. 07 asics nimbus 18 104 2. 88% 33. 44 118. 08 asics gel nimbus 18 130 2. 31% 44. 21 116. 98 asics gel cumulus 17 women 84 0. 00% 43. 37 115. 45 … #ecommercial 16 @whoop_app

Scored terms based on an advanced model (Potential) Low performers - Last 30 Days Clicks CR Cost Score knee bandage 13 0. 00% 5. 23 1. 5 asics 25 0. 00% 12. 89 1. 67 sports duffle bag 17 0. 00% 4. 29 2. 11 trail shoes 16 0. 00% 6. 82 2. 25 beach slippers 33 0. 00% 16. 86 2. 53 adidas leisure suit 13 0. 00% 4. 27 3. 53 nike duffle bag 13 0. 00% 3. 63 3. 79 nike bag 19 0. 00% 6. 11 3. 85 adidas running shoes 28 0. 00% 16. 46 3. 89 nike slippers 14 0. 00% 5. 28 3. 89 running shoes women 13 0. 00% 6. 23 3. 94 … #ecommercial 16 @whoop_app

Implementation guide Low Performers List high (lower funnel) Add with medium Priority [knee bandage] [sports duffle bag] [trail shoes] [beach slippers] [adidas leisure suit] as negative Exact flow Low Performers List High Performers List Add 2 Campaigns per existing Shopping Campaign Mid (mid funnel) flow Existing Campaign Set to High Priority “asics gt 2000 4” “adidas zx 750” “asics gel cumulus 17” “nike metcon 2” “asics kayano 22” as negative Phrase [knee bandage] [sports duffle bag] [trail shoes] [beach slippers] [adidas leisure suit] as negative Exact low (upper funnel) Add with low priority #ecommercial 16 All non-polarized queries are managed via the mid funnel which serves as a pool for them. @whoop_app

Choose the right Campaign Priorities Don’t’ forget to clone all other campaign settings #ecommercial 16 @whoop_app

Tip: Use Negative Keyword Lists And add them to all applicable campaigns. #ecommercial 16 @whoop_app

The Upper Funnel Exact List to Mid and Lower Funnel And add them to all applicable campaigns. #ecommercial 16 @whoop_app

And the Lower Funnel Phrase only to the Mid Funnel And add them to all applicable campaigns. #ecommercial 16 @whoop_app

Analyzing the Attribution Behaviour of Google Shopping Interest Comparison Decision #ecommercial 16 Branded Search @whoop_app

Analyzing the Attribution Behaviour of Google Shopping After some weeks you are able to get better insights #ecommercial 16 @whoop_app

New: Data Driven Attribution (DDA) & Integration in Conversions 1. If you have over 800 conversions (suggested by Google) you can use DDA 2. You can apply an attribution model to your conversion measurement #ecommercial 16 @whoop_app

What does that mean for Shopping/PLAs? You can shed some light on attribution behavior of queries and products click “running shoes” “nike running shoes” “buy nike free 5. 0 shoes” Conversion: 100 £ TODAY 0 £ 100 £ last-click wins FUTURE 33. 33 £ e. g. linear model Important note: Automated Bidding for Google Shopping will continue to use the last-click-model #ecommercial 16 @whoop_app

THANK YOU! SEE YOU AT THE NEXT #ecommercial 16 @whoop_app