0% found this document useful (0 votes)
80 views

Search Advertising Driven by Machine Learning: Michaela Woloszczak Denis Dautaj

This document discusses how machine learning can help with search advertising. It explains that machine learning helps automate tasks like bidding, targeting, formats and attribution. It promotes using smart bidding which leverages exclusive signals like location, device, language and search queries to determine the most relevant bids at auction time. The document recommends using machine learning to focus on high value tasks and outlines different automated bidding options in Google Ads.

Uploaded by

Nernardo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
80 views

Search Advertising Driven by Machine Learning: Michaela Woloszczak Denis Dautaj

This document discusses how machine learning can help with search advertising. It explains that machine learning helps automate tasks like bidding, targeting, formats and attribution. It promotes using smart bidding which leverages exclusive signals like location, device, language and search queries to determine the most relevant bids at auction time. The document recommends using machine learning to focus on high value tasks and outlines different automated bidding options in Google Ads.

Uploaded by

Nernardo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 48

Search Advertising driven by

Machine Learning
Michaela Woloszczak Denis Dautaj
Search Audience & Formats, Search Automation,
Central Europe Central Europe

Confidential + Proprietary Confidential + Proprietary


1. How Machine Learning can help you

2. Automation framework to success

3. Key takeaways to grow your business

Confidential + Proprietary
1. How Machine Learning can help you

2. Automation framework to success

3. Key takeaways to grow your business

Confidential + Proprietary
We are going from a
“mobile first world”
to an “AI first world”

Confidential + Proprietary
Machine Learning helps us in our private lives already

Confidential + Proprietary
Machine Learning helps us in advertising, too

Bidding Targeting Formats Attribution

Confidential + Proprietary
Why do we need ML for advertising?

Searches for Visits Researches


singlespeed online bike on bike blogs
bike on mobile store on laptop

Uses voice
Buys
search to Re-visits singlespeed
search for online bike bike at local
unique store store
singlespeed
bikes

Confidential + Proprietary
Why do we need ML for advertising?
Operating
Time of System
Week Version

Actual Ad User
Copy Conversion Conversation Cross Device Operating
Value Rate Conversions System

Keyword Audience Ads Location Time Device

Actual
Language
Search RLSA Mobile App
Setting Browser
Query Device Bids
Type
Similar
Audience
Tablet
DFSA Device

Confidential + Proprietary
Machine Learning can help us focus on the things where
we can add value
Today Tomorrow

Reporting Reporting
UX
UX

Creatives
(Content)
Creatives Attribution
(Formats) Bidding Bidding
Attribution

Targeting Targeting Measurement


(KWs, Measurement (Conceptual)
Audience)

BCG: Cutting Complexity, Adding Value (2013): we spend 80% of our time on manual, low-value tasks
Bubble size representing the avg. amount of time spent on certain tasks

Confidential + Proprietary
1. How Machine Learning can help you

2. Automation framework to success

3. Key takeaways to grow your business

Confidential + Proprietary
1 Leverage
Smart Bidding 2 Feed in all
Audience Signals 3 Show most
relevant Creatives

Confidential + Proprietary
1 Leverage
Smart Bidding 2 Feed in all
Audience Signals 3 Show most
relevant Creatives

Confidential + Proprietary
There are many automated bidding options on the market

AdWords

Automated
DoubleClick
Rule
based
Manual

Other
Tools
Automation

Confidential + Proprietary
True auction-time bidding – powered by exclusive signals
User-specific bids Combination
Exclusive signals
through auction-time bidding of signals

Time
OS Browser

Geo Device Language Search Query

BID
Ad Format Search Partners

Remarketing Lists

Only available with


Adwords Smart Bidding
Confidential + Proprietary
How do you deal with these signals?
Performance-Difference to avg.
(Conv. Rate)

● User in Berlin -40 %

Geo
● User in Hamburg +60 %

-35 %
● User on Saturday
+45 %
Time
● User on Sunday
Your
-20 %
● User on Smartphone
● User on Desktop
+35 % bids?
Device

-20 %
● User in Berlin, on Saturday, with Android
● User in Berlin, on Saturday, with iPhone +80 %

∞ Confidential + Proprietary
Get the most of your campaigns with query-level bidding

Query*

Keyword buy jeans

jeans jeans

wash jeans

Keywords (Broad, Phrase) Dynamic Search Ads (DSA) Google Shopping

Manual: keyword-level Manual: category-level Manual: product-level

Smart Bidding: query-level Smart Bidding: query-level Smart Bidding: query-level

*Note: 15% of queries on Google are new every day Confidential + Proprietary
AdWords Smart Bidding options, aligned to your goals
Campaign Goal Recommended Strategy

Target Outranking Share


VISIBILITY
No Target Search Page Location
conversion
tracking
WEBSITE CLICK Maximize Clicks

Basic eCPC (semi-automated)


CONVERSIONS / SALES
Advanced tCPA

Excellence $ REVENUE tROAS

Confidential + Proprietary
Newest features for even stronger results

Better target More predictive


accuracy signals

Confidential + Proprietary
Transparency and control via the shared library
Monitor the performance of
flexible bid strategies key KPIs
and watch how it improves
over time

Understand the strategy status


and potential limitations, then
take action

Confidential + Proprietary
Your road to success with a clean pre/post test
1. Opt-in Target CPA Bid Strategy 2. Pre-Post evaluation
CPA Target = 30-day CPA avg. (Ignore “learning” week and
Minimum 30 conversions most recent week)

Control Period Learning Run Test Conversion Evaluate /


2+ weeks 1+ week 2-4 weeks Delay Adjust
~1 week

Ignore in Check campaign Ignore in


evaluation! status column evaluation!

3. Potential target adjustment


(within +/- 20%)

Confidential + Proprietary
Or via A/B testing with Drafts & Experiments

Enter a specific date range

Determine the portion of traffic


you want to go through the
experiment

Assess performance

Confidential + Proprietary
Finding suitable campaigns for a test
Recommended Recommended
Campaign Goal
Strategy Conversions*
Target: Outranking Share
VISIBILITY
Search Page Location

WEBSITE CLICK Maximize Clicks

eCPC (semi-automated) -
CONVERSIONS / SALES
tCPA >30

$ REVENUE tROAS >50

* per campaign or portfolio in the last 30 days;


NOTE: Target ROAS requires 15 conversions in past 30 days technically Confidential + Proprietary
We can help you identify campaigns for a tCPA test
We create estimates using your account’s Bid Simulator data.
Bid Simulator looks at the specific auctions your accounts participated in during a
recent past week to estimate these performance gains.
100 +30 +30% +0%
Campaign 1 CURRENT CONVERSIONS SIMULATED CONVERSION SIMULATED CONVERSION SIMULATED CPA CHANGE
UPLIFT UPLIFT %

82 +19 +23% +0%


Campaign 2 CURRENT CONVERSIONS SIMULATED CONVERSION SIMULATED CONVERSION SIMULATED CPA CHANGE
UPLIFT UPLIFT %

... ... ... ... ...

Note: We generate traffic estimates based on historical performance. These are not a guarantee of future performance as the context/competition
etc. will be different from the past. Confidential + Proprietary
1 Leverage
Smart Bidding 2 Feed in all
Audience Signals 3 Show most
relevant Creatives

Confidential + Proprietary
Audience is a key signal for Smart Bidding

singlespeed bikes

Time OS Browser

Visited product detail


site 5 days ago
Device Language Search Query
Audience

Geo Ad Format Search Partners

bid down bid up

Confidential + Proprietary
The Search Audience suite has grown over the years

Your Data
Remarketing Lists for Search Ads (RLSA) – (2013)

+
Customer Match (2015)
singlespeed bikes

Google Data
Similar Audiences (2017)
Demographics for Search Ads (2016)

Confidential + Proprietary
Smart Bidding now uses even richer Audience Signals
Performance-Difference to avg. Your bid?
(Conv. Rate)

● User new to website -20 %


● User on All Visitors list +110 %
Audience List
Membership
New

● User in Shopping Cart 76 days ago -35 %


● User in Shopping Cart 1 day ago +45 %
Visitor Recency

● Men -40 %
Demographics ● Women +40 %

Confidential + Proprietary
How to feed in Audience Signals for maximum success

1 2
Target all Apply to
relevant users all search
campaigns
(including DSA &
Shopping)
with bid only &
campaign-level

Confidential + Proprietary
Segment Search Audience for success

General Visitors

Category

Product

Cart Abandoner

Buyer

Remarketing Lists for Similar Audience


Search Ads

Confidential + Proprietary
How to feed in Audience Signals for maximum success

1 2
Target all Apply to
relevant users all search
campaigns
(including DSA &
Shopping)
with bid only &
campaign-level

Confidential + Proprietary
Simplify management with Campaign-Level Audiences
Before After
List 1
AdGroup List 2 AdGroup
List 3
List 1 List 1
Campaign AdGroup List 2 Campaign List 2 AdGroup
List 3 List 3
List 1
AdGroup List 2 AdGroup
List 3

Easier ✔ More data for Reduced


✔ ✔
management better bids criteria limits

Confidential + Proprietary
Really using all audience signals for Smart Bidding with
Data-driven Attribution

Data-driven
model

Cross-device ✔ More data for



touch-points smarter bids

Confidential + Proprietary
1 Leverage
Smart Bidding 2 Feed in all
Audience Signals 3 Show most
relevant Creatives

Confidential + Proprietary
Which one is the best ad?
CTR: 8%
a CR: 5%
CPA: €20

CTR: 10%
b CR: 2%
CPA: €22

CTR: 12%
c CR: 1%
CPA: €40

Confidential + Proprietary
Each search has a different context

singlespeed bikes

3pm Android Chrome

Samsung English Hamburg

Confidential + Proprietary
Each search has a different context

singlespeed bikes

8pm iOS Safari

iPhone English Munich

Confidential + Proprietary
Each search has a different context

singlespeed bikes

8am Apple Safari

Laptop German Berlin

Confidential + Proprietary
Leverage Machine Learning to show the best ad
for every query

Ad groups with 3 or more


high-quality ads may receive
up to 15% more clicks*,
compared to ad groups with
only 1 or 2 ads.

*when running on optimized ad rotation

Confidential + Proprietary
Personalize ads at scale with IF Functions

singlespeed bikes

Single Speed Bike - Great Range


www.bike-example.com/Sale
{=IF(audience IN (Converters - 540 days), Come
back for 60% off): Free shipping on your first
order}. Next day delivery guaranteed.

Confidential + Proprietary
1 Leverage
Smart Bidding 2 Feed in all
Audience Signals 3 Show most
relevant Creatives

Account Set Up

Confidential + Proprietary
Simplify your accounts for an AI first world
Manual Bidding Smart Bidding
Example of granular, duplicated campaigns based on Consolidated campaign structure supercharging
matchtype, device, RLSA and DSA for manual bidding. Smart Bidding with broad match types and DSA.
adgroup
Broad Match Campaign adgroup
adgroup
adgroup
matchtype Exact Match Campaign adgroup
adgroup
adgroup
Phrase Match Campaign adgroup
adgroup Campaign Broad match adgroup
adgroup Devices / Audiences / Geo / Time
Mobile Campaign adgroup
adgroup
adgroup
device Desktop Campaign adgroup
adgroup
adgroup Or
Tablet Campaign adgroup
adgroup

adgroup
10 day (RLSA) Campaign adgroup
adgroup Exact match adgroup
adgroup
RLSA 20 day (RLSA) Campaign adgroup Campaign
adgroup
Devices / Audiences / Geo / Time DSA adgroup
adgroup
30 day (RLSA) Campaign adgroup
adgroup

adgroup
DSA Dynamic Search Ads adgroup
adgroup

Confidential + Proprietary
And utilize scripts for quick and easy implementation

Link to AdWords scripts library


Confidential + Proprietary
1. How Machine Learning can help you

2. Automation framework to success

3. Key takeaways to grow your business

Confidential + Proprietary
5 key takeaways to grow your business

01 Automation allows us to shift time to important tasks and grow your business

02 Leverage Smart Bidding to handle infinite data signals for user-specific bids

03 Incorporate audience data everywhere to make Smart Bidding even smarter

04 Automate your creatives for most relevant messaging

05 Simplify your account setup and utilize scripts for more efficient implementation

Confidential + Proprietary
Thank you!
Michaela Woloszczak | Denis Dautaj

Slides & Feedback


https://goo.gl/forms/1ariaYsZmQv17X7H3

Confidential + Proprietary Confidential + Proprietary


APPENDIX

Confidential + Proprietary Confidential + Proprietary


tCPA worked well for CosmosDirekt
“Durch Smart Bidding sind wir unabhängiger, weil der Automatisierungsgrad gestiegen ist. Wir schaffen
es nun überwiegend Inhouse zu arbeiten und uns auf strategische und taktische Themen zu
konzentrieren, als uns mit operativer Kampagnenarbeit aufzuhalten.”

- Dennis Brill, Online Marketing Manager SEA bei CosmosDirekt

Mobile CPA Share of Mobile Conversions


- 5% budget now 25 % + 100%

Confidential + Proprietary
Simplify your Search Audience Strategy for success
Many accounts look like this... This is what an excellent setup looks like
Running separate RLSA campaigns as target & bid Leveraging Search Audience signals in every search
campaign with bid only and at campaign-level

540 day RLSA: All website


10 days Homepage excluded
30 days Homepage excluded
540 day RLSA: Homepage
Original 10 days Cart excluded
Campaign
AdGroup 1 30 days Cart excluded 540 day RLSA: Category
10 days Converter excluded
30 days Converter excluded 540 day RLSA: Product
Campaign
540 day RLSA: Cart
w/ Smart Bidding
10 days Homepage targeted 540 day RLSA: Sale
30 days Homepage targeted
RLSA 10 days Cart targeted Similar Audience: Sale
Campaign AdGroup 1 30 days Cart targeted
10 days Converter targeted Similar Audience: Cart
30 days Converter targeted
Similar Audience: Product

Confidential + Proprietary

You might also like