In this deck you will learn about various tactics for negative keyword automation to tackle the changing exact match and increasing query overlaps across adgroups. The talk was held at MSX Munich 2019 by Christopher Gutknecht, Head of Online Marketing at norisk.
The accompanying scripts can be found at: bit.ly/norisk_negatives
Enjoy the content, in case of any questions, reach out via linkedin:
https://www.linkedin.com/in/chrisgutknecht/
4. Today’s 7-Part Negative Framework
DSA SEARCH
2. DSA List-Fencing
SHOPPING
1. Negative-List Sync 3. Query Overlap
4. Near Exact Add
5. Negative Conflicts
6. Negative Deduplication in Ads Editor
7. Bad NGram Finder with Big Query
ALL
5. I’m a PPC With a Passion For Automation
Javascript
Python
2008 2014 2016
PPC Automation
PPC
20182012
Christopher Gutknecht
Head of Online Marketing @ norisk
Munich-based
Focus Ecom & Retail
Self Taught Dev
Dad of 2,5 yr old
6. My Session Material For You
SCRIPTSSLIDES
slideshare.net/norisk bit.ly/norisk_negatives
17. Common Pitfall: Over & Underfencing
1. Negative-List Sync
OVERFENCING UNDERFENCING
Categories “Laufschuhe Stadt & Wald”
“aalborg GmbH & Co KG”
→ Regularly Check Your Input Data!
“Lo”
Brands “On”
18. Addon: Exact Match Shopping Campaigns
1. Negative-List Sync
EXACT MATCH SHOPPING FENCER
Brand_Exact
Query Negative List
Brands_Broad
Script Download: bit.ly/norisk_negatives
Query
Set Negative if not equals
Campaign
25. 2. Search Queries
Volume-based Fencing Script For DSA
2. DSA List-Fencing
DSA_Fencing_Ex_1
DSA_Fencing_Ex_2
DSA_Fencing_Ex_3
....
{max = 5}
1. DSA Queries
DSA List Pool
DSA-FENCING PROCESS (EXACT)
if (Conv=0 & Overlap)
checkOverlap
if (clicks > x)
Script Download: bit.ly/norisk_negatives
Sync-Script Sources
26. 3. Handling Query Overlaps in Search
DSA SEARCHSHOPPING
3. Query Overlap
4. Near Exact Add
5. Negative Conflicts
ALL
27. Misrouted Queries Are Expensive
3. Query Overlap
Query Overlaps = Hidden Cost Drivers!
0.07 vs 0.32 € → 3.5 times more expensive
28. Solving Overlaps by Word Similarity
3. Query Overlaps
+Nike
“Nike Air Max”
Nike Air Max 97
Keyword LetterChanges
(Levensthein)
Similarity
Nike 9 27%
0 (/ 15) 1
Nike Air Max 3 80%
QUERY “NIKE AIR MAX 97“
29. Suggesting Negative For Broader KW
3. Query Overlaps
+Nike
“Nike Air Max”
Nike Air Max 97
Keyword LetterChanges Similarity
Nike 9 27%
0 1
Nike Air Max 3 80%
QUERY “NIKE AIR MAX 97“
“Air Max 97”
35. 4. Handling Expensive Near-Exact
DSA SEARCHSHOPPING
3. Query Overlap
4. Near Exact Add
5. Negative Conflicts
ALL
36. What To Do With Exact Variants?
4. Near-Exact Split
CHOOSE BY TYPE, CLICKS AND CPC
Word Order
Word Variant
Spacing Variant
Typo
Letter Variant
37. How To Distinguish Typos & Variants?
4. Near-Exact Split
GOOGLE SUGGEST & CUSTOM SEARCH
1. Suggest
for Variants
2. Custom
Search for
Typos
38. The Uncertainty With Similar Keywords
4. Near-Exact Split
SIMILAR KWS IN THE [ SAME ] AD GROUP
39. Near-Exact in 2019: End of the SKAG?
4. Near-Exact Split
SKAGS IN 2019 : PROS & CONS
Source: unbounce.com/ppc/skags-ppc-best-practice/
High CTR & Quality Score
Query Overlaps if Near-Exact
40. Near-Exact Handler Script: What It Does
4. Near-Exact Split
ADDS KEYWORDS OR CREATES NEW ADGROUPS
Script Download: bit.ly/norisk_negatives
1. Add Keyword
Queries
Query
wordSimilarity = high
Actions
2. New SKAG AdGroupQuery 2
Query 3 3. All Typo AdGroup
clicks greater x
is Typo = true
41. 5. Resolving Negative Conflicts
DSA SEARCHSHOPPING
3. Query Overlap
4. Near Exact Add
5. Negative Conflicts
ALL
42. Google’s Negative Conflicts MCC Script
5. Negative Conflicts
SCRIPT OUTPUT IN SHEET
Run Weekly → Will NOT (!) cover DSA & Shopping
43. How To Find Harmful Negatives? (Hard)
5. Negative Conflicts
LOOK FOR LONG-TERM IMPRESSION DROPS
Impression Drops without Status or CPC Changes
45. Use the Deduplication Tool in Ads Editor
6. Neg Deduplication
Set Scope &
Configuration
> Tools
46. Use the Deduplication Tool in Ads Editor
Set Scope & Configuration
6. Neg Deduplication
ALL
47. 7. Finding “Bad“ Ngrams with Big Query
DSA SEARCHSHOPPING
7. Bad NGram Finder with Big Query
ALL
48. Ngramming Is Better with BigQuery
7. Find Bad Ngrams
BRAINLABS SCRIPT > TIMEOUTS
49. The Basis: Google Ads Data Transfer
7. Find Bad Ngrams
ALL SETTING UP THE ADS TRANSFER: 5 MIN
Set up Cloud console & go to: bigquery.cloud.google.com
50. Where To Run Ngram Queries?
7. Find Bad Ngrams
BIG QUERY GOOGLE COLAB DATA STUDIO
51. Pre-Analysis: Data by Word Count
7. Find Bad Ngrams
QUERY PERFORMANCE BY WORD COUNT
No relevant long tail here!
52. Example Unigrams: Not Helpful
7. Find Bad Ngrams
1-WORD RESULTS (UNIGRAMS)
No clear indication of “bad” words
53. Example Bigrams: Not Helpful
7. Find Bad Ngrams
2-WORD RESULTS (BIGRAMS)
No clear indication of “bad” words
54. RECAP: 7-Part Negative Framework
DSA SEARCH
2. DSA List-Fencing
SHOPPING
1. Negative-List Sync 3. Query Overlap
4. Near Exact Add
5. Negative Conflicts
6. Negative Deduplication in Ads Editor
7. Bad NGram Finder with Big Query
ALL
55. RECAP: My Session Material For You
SCRIPTSSLIDES
slideshare.net/norisk bit.ly/norisk_negatives
56. Takeaways: Take Control of Negatives!
1. Face your overlaps by volume
2. Use the 7-part framework
3. Never assume all is ok
4. Save € & reinvest WE’RE HIRING!
#SMXMunich