Google Ads Audience Manager for Media Planners and Buyers

 

How to Build Audience Strategy: A Practical 101 Guide Using a Realistic E-commerce Scenario

Who This Is For (And Who It Is Not)

This is for:

→ media planners and buyers owning budget, structure, and performance across Search, Display, YouTube, and Performance Max
→ performance marketers responsible for CAC, ROAS, and LTV outcomes
→ operators managing scale, not just campaigns

This is NOT for:

→ UI walkthrough seekers
→ theory-only readers
→ anyone not accountable for revenue outcomes

 

Introduction

Google Ads does not become inefficient because of competition alone.

It becomes inefficient when:

→ the system learns from the wrong users
→ intent layers are mixed
→ high-value and low-value users are treated the same
→ exclusions are weak or missing

That leads to:

→ rising CAC
→ unstable ROAS
→ wasted remarketing spend
→ poor scaling

Audience Manager is not just a feature.

It is the input system that defines how the entire account learns.

If the inputs are weak:

→ every campaign suffers

If the inputs are strong:

→ every campaign improves

 

A Simple Mental Model

Audience Manager is not about “audiences”.

It is about signal engineering:

→ data source
→ event quality
→ segmentation logic
→ value separation
→ campaign usage

Each layer feeds the next.

Break one layer, and performance drops.

 

How Data Flows (Realistically)

This is how a real e-commerce account operates:

→ user enters via Search, Display, YouTube, or direct
→ GA4 / GTM captures actions (view_item, add_to_cart, purchase)
→ those events populate segments in Audience Manager
→ segments are applied differently across campaigns

→ Search reads them as signals
→ Display uses them for control and remarketing
→ YouTube uses them for demand generation
→ Performance Max uses them as expansion inputs

→ conversions feed back into Smart Bidding
→ system optimizes based on who converts and what value they generate

Key point:

→ Audience Manager does not just influence targeting
→ it defines learning direction

 

The Real Campaign Context: Summer E-commerce Push

We are working with a consumer e-commerce brand in the hydration and wellness category.

Product ecosystem:

→ hydration powders and electrolyte mixes
→ fitness-focused products
→ travel sachets and starter kits
→ bundles designed to increase AOV

Market reality in summer:

→ demand spikes
→ CPCs increase
→ competition intensifies
→ new users enter the category

Business targets:

→ +35% revenue
→ 4.5x ROAS
→ −20% CAC
→ 60% new customers
→ +15% AOV

At this stage, the problem is not reach.

The problem is:

→ separating high-intent vs low-intent users fast enough
→ prioritizing high-value users correctly

 

Audience Lifecycle (How Users Move)

Stop thinking funnel. Think conversion probability curve.

→ cold user → no signal
→ site visitor → weak signal
→ product viewer → intent forming
→ cart user → high intent
→ buyer → confirmed value
→ repeat buyer → profit driver

Each stage has:

→ different CVR
→ different AOV
→ different bid ceiling

Example:

→ cart users convert at significantly higher rates than product viewers
→ repeat buyers often deliver higher AOV

If these users are not separated:

→ bidding becomes inaccurate
→ CAC increases
→ scaling becomes inefficient

 

What Audience Manager Actually Is

Inside Google Ads:

Audience Manager is a rule-based segmentation system.

You are defining:

→ who enters which segment
→ based on which behavior
→ for how long
→ with what exclusions

It includes:

→ Your data segments (first-party)
→ Custom segments (intent-based)
→ Combined segments (persona logic)
→ Your data insights (analysis layer)

It is not about targeting.

It is about:

→ controlling who the system learns from

 

Where Audience Manager Sits (Platform Context)

→ centralized in shared library
→ connected to Data Manager
→ feeds all campaigns

Meaning:

→ weak structure impacts entire account
→ strong structure improves entire account

 

Targeting vs Observation vs Signals

Each campaign type uses audiences differently:

→ Search → observation → reads signals, adjusts bids
→ Display → targeting → controls delivery
→ YouTube → audience-led → generates demand
→ Performance Max → signals → expands reach

Critical nuance:

→ audiences behave differently depending on where they are used

 

How to Build Audience Strategy Using Actual Options Available

Step 1: Start from revenue logic, not segments

Ask:

→ who converts fastest
→ who drives highest value
→ who should be excluded

 

Step 2: Build event hierarchy

Core events:

→ view_item
→ add_to_cart
→ begin_checkout
→ purchase

Advanced layer:

→ high-value purchase vs low-value
→ repeat vs first-time

Without this:

→ system optimizes for volume, not value

 

Step 3: Segment by intent AND value

Do not stop at:

→ cart users

Break it further:

→ cart users (high value)
→ cart users (low value)
→ repeat cart users

Now you control:

→ bids
→ budgets
→ signals

 

Step 4: Apply strict exclusion logic

Examples:

→ exclude buyers from acquisition
→ exclude cart users from product viewers
→ exclude recent buyers from remarketing

Impact:

→ cleaner signals
→ reduced waste
→ better CAC

 

Step 5: Use time windows as intent decay

→ 0–3 days → highest urgency
→ 4–7 days → strong intent
→ 14–30 days → declining intent
→ 60–90 days → reactivation

 

Step 6: Use customer lists for value injection

→ high AOV users
→ repeat buyers
→ frequent purchasers

Used for:

→ scaling
→ upsell
→ value-based optimization

 

Step 7: Control overlap

Same user can exist in multiple segments.

Without control:

→ conflicting signals
→ inefficient spend

Solution:

→ clear hierarchy
→ exclusions

 

Step 8: Align with campaigns

→ Search → signal reading
→ Display → conversion control
→ YouTube → demand creation
→ Performance Max → scaling

 

Step 9: Use audience insights

→ identify high-performing segments
→ adjust bids
→ expand targeting

 

Step 10: Iterate continuously

→ test
→ learn
→ refine
→ scale

 

The Complete Audience Landscape (Aligned to Actual Platform Options)

Audience Creation vs Audience Usage

Creation:

→ defining rules

Usage:

→ applying in campaigns

Most failure happens here.

 

Audience Size, Overlap, and Learning

→ small = no delivery
→ large = weak signals
→ overlap = confusion

 

Data Freshness and Recency

→ recent users = high probability
→ older users = lower probability

 

Privacy and Signal Strength

→ tracking loss is real

Solution:

→ Enhanced Conversions
→ first-party data
→ CRM enrichment

 

Audience Insights in Action

→ validate assumptions
→ find new segments
→ optimize spend

 

Structuring Audience Groups

→ high intent → cart
→ mid intent → product
→ low intent → cold
→ value → repeat buyers

 

Creative Alignment by Audience

→ cart → urgency + incentive
→ product → differentiation
→ cold → education
→ repeat → bundles

 

Budget Allocation Logic

This is where media planning decisions directly impact revenue.

→ cart users → highest budget allocation because they have the highest conversion probability and shortest path to revenue

→ product viewers → controlled, mid-level budget because they are still evaluating and require persuasion before scaling spend

→ cold audiences → limited, test-driven budget because they are essential for discovery but deliver the lowest immediate return

If this is inverted:

→ CAC increases rapidly
→ remarketing underperforms
→ scaling becomes unstable

 

How This Drives Results

→ clean segmentation improves signal quality
→ strong signals improve bidding
→ better bidding reduces CAC

 

Where Most Strategies Break

→ no value segmentation
→ weak exclusions
→ over-reliance on generic audiences
→ poor signal quality

 

Execution Checklist (First 30 Days)

Week 1:

→ validate events + build segments

Week 2:

→ map segments to campaigns

Week 3:

→ analyze performance

Week 4:

→ scale high-value segments

 

What a Well-Structured Account Looks Like

→ clear intent separation
→ value-based segmentation
→ minimal overlap
→ strong first-party signals

 

Final Thought

Audience Manager is not about audiences.

It is about:

→ controlling the learning system

That is what separates:

→ efficient scale
→ from expensive growth

 

How are you structuring your audience signals across your account today?

→ are you separating users by intent and value
→ are you controlling exclusions properly
→ or are you still feeding mixed signals into the system and expecting stable performance

 

The 2026 Programmatic Reality: Why Media Buying Performance Is Now Driven by Structure, Not Just Bidding in Display & Video 360 (DV360)

 

Having managed programmatic budgets on both the brand and agency side, I’ve seen where performance actually breaks. In 2026, it’s not the algorithm. It’s how you structure and control it.

Earlier, the playbook in Display & Video 360 (DV360) was simple:
→ build campaigns
→ let the platform optimise
→ scale

Now, that approach doesn’t hold up.

Performance today isn’t defined by how much you automate inside DV360.
It’s defined by how well you control the inputs that drive that automation.

Most inefficiencies in DV360 campaigns don’t come from bidding. They come from structure, duplicated reach, and poor supply decisions.

Let’s break this down using one real scenario.

 

The Scenario: Summer Dresses Campaign in Display & Video 360 (DV360)

An eCommerce brand is running a summer campaign for dresses, managed by an agency.

The goal:

• drive revenue
• maintain strong ROAS
• scale without increasing CPA

This campaign is executed inside Display & Video 360.

 

How This Campaign Is Built in Display & Video 360 (DV360)

Partner (Agency Level in DV360)

• Controls brand safety and inventory exclusions

Example:

• Blocks MFA websites and low-quality mobile apps
• Excludes irrelevant categories at scale

👉 Business impact: prevents wasted spend before DV360 even enters auctions

 

Advertiser (Brand Level in DV360)

• Holds creatives, audiences, and Floodlight activities

Example:

• Creatives: summer dress banners and video ads
• Audiences:
• users who viewed dresses
• cart abandoners
• Floodlight tracking:
• add-to-cart
• purchase

👉 Business impact: ensures accurate conversion tracking and ROAS measurement inside DV360

 

Campaign (DV360 Campaign Level)

• Groups related line items under a defined campaign

Example:

• “Summer Dresses – Q2 Campaign”

👉 Business impact: supports reporting and setup clarity, not optimisation

 

Insertion Order (IO in DV360)

• Controls:
• budget
• pacing
• flight dates

Example:

• IO 1: Prospecting (€50K)
• IO 2: Remarketing (€20K)

👉 Business impact: controls how budget is allocated and spent within DV360

 

Line Items (Core Buying Unit in DV360)

• Controls:
• targeting (audience, geo, device, inventory)
• bidding strategy
• optimisation signals

Example:

• In-market fashion audience (Display)
• Contextual fashion inventory
• CTV video targeting
• Remarketing (dress viewers)

👉 Business impact: directly drives:

• CPA
• ROAS
• revenue

 

👉 Simple DV360 logic:

• Insertion Order = how much you spend
• Line Item = how that spend generates revenue

This is where the 2026 shift becomes critical.

 

Campaign Structure: Where ROAS Is Actually Won or Lost in DV360

Before (Typical DV360 Setup)

For this dresses campaign:

• Line Item 1 → In-market fashion audience
• Line Item 2 → Broad lifestyle audience
• Line Item 3 → Contextual fashion targeting

Now imagine a DV360 user who:

• is browsing fashion content
• is also in-market for dresses

👉 That same user qualifies for all 3 Line Items

What happens inside DV360?

• All 3 Line Items enter the auction
• DV360 bids through multiple line items
• You compete against yourself

👉 Business impact:

• higher CPMs
• higher CPA
• lower ROAS

 

Now (Optimised DV360 Structure)

• Line Item 1 → Broad prospecting
• Line Item 2 → High-intent audience (in-market)
• Line Item 3 → Remarketing (fully isolated)

With:

• no overlap
• clear role per Line Item

👉 What changes:

• cleaner signals inside DV360
• faster optimisation
• more stable delivery

👉 Business impact:

• lower CPA
• higher ROAS
• better scalability

 

The Strategic Fix

• Each DV360 Line Item has one clear job
• No duplication in targeting
• Insertion Orders handle only budget
• Partner level handles exclusions

This also avoids bid shading inefficiencies, where multiple DV360 line items targeting the same user force the platform to split or mis-prioritise bids.

This is what I call signal governance: controlling the inputs that drive DV360 decision-making.

 

Duplicate Reach: The Hidden Reason DV360 CPA Increases

Before

A user views a dress on the site.

Then through DV360:

• sees a display ad on mobile
• later sees a CTV ad
• then sees another display ad on desktop

👉 Same DV360 user, multiple impressions

What happens?

• frequency increases
• conversions don’t increase

👉 Example:

• 1 user sees 8–10 impressions
• but converts once

👉 Business impact:

• wasted impressions
• rising CPMs
• no incremental revenue

 

Now

We actively manage reach inside DV360.

• Identify overlap across:
• CTV
• mobile
• desktop
• Reduce spend where frequency is too high
• Shift budget to new users

👉 Example:

• reduce CTV frequency from 8 to 3
• reallocate budget to new prospecting users

👉 Business impact:

• more unique users reached
• better CPA
• improved efficiency

 

The Strategic Fix

• Analyse overlap using DV360 reporting
• Reduce redundant impressions
• Reinvest into incremental reach

 

Where Your DV360 Ads Run Matters More Than Your CPM

Before

• Heavy reliance on open exchange in DV360
• Focus on lowering CPM

👉 What happens:

• ads show on low-quality or MFA inventory
• users don’t engage

👉 Business impact:

• poor conversion rates
• weak ROAS

 

Now

For the same DV360 campaign:

• Shift budget to:
• curated deals
• premium publishers
• high-quality CTV inventory
• Apply Supply Path Optimisation (SPO)

👉 Example:

• reduce open exchange spend by 30%
• move budget to curated fashion publishers

👉 What changes:

• better placements
• higher engagement
• improved conversions

👉 Business impact:

• higher ROAS
• stronger revenue performance

 

The Strategic Fix

• prioritise quality over cheap CPM
• buy through cleaner supply paths
• optimise for outcomes, not cost

 

What This Means for Media Planning and Buying in DV360

In 2026, DV360 executes the bidding, but the human defines the architecture.

Success doesn’t come from:

• adding more targeting
• increasing automation

It comes from:

• clean DV360 structure
• controlled reach
• efficient supply paths

Because these directly impact:

• ROAS
• CPA
• Revenue growth
• Scalability

This is signal governance in DV360.

 

Final Takeaway for Media Buyers

Media efficiency in Display & Video 360 (DV360) is not won in the bidding algorithm.

It is won in:

• how you structure Line Items
• how you control reach and frequency
• how you optimise your supply path

If you’re not actively reviewing:

• Line Item overlap
• reach distribution
• supply paths

then part of your DV360 budget is being wasted without visibility.

And in a performance-driven environment, that is exactly where the real advantage lies.

 

Cohort Analysis for Performance Marketing and Advertising Campaigns

 

In performance marketing, most reporting is built around immediate outcomes such as conversions, CPA, and ROAS. These metrics are important because they show whether campaigns are generating demand efficiently. But they only describe what happens at the point of acquisition or conversion.

They do not tell you what happens next.

Do those newly acquired customers come back and buy again? Do they generate meaningful revenue over time? Are newer campaigns bringing in better customers than older ones, or just cheaper ones? Are recent optimizations actually improving long-term business value, or only making short-term numbers look better?

This is where cohort analysis becomes one of the most useful tools in performance marketing.

Cohort analysis helps marketers move from snapshot reporting to lifecycle understanding. Instead of treating all customers as one blended mass, it groups them by a shared starting point and tracks what happens to each group over time. That is what makes it so powerful for advertising, retention, budgeting, creative strategy, and growth planning.

What Cohort Analysis Actually Means

A cohort is a group of users or customers who begin their journey at the same time or share the same starting event.

In performance marketing, the most common version is an acquisition cohort:
customers who made their first purchase, first signup, or first conversion in the same day, week, or month.

So if you are running an e-commerce business, your cohorts might look like this:

• Customers whose first purchase happened in Week 1
• Customers whose first purchase happened in Week 2
• Customers whose first purchase happened in Week 3

The point is not just to know how many customers each week brought in. The point is to see what each of those groups did afterward.

Did they buy again?
Did they spend more?
Did they disappear?
Did newer cohorts perform better than older ones?

That is the heart of cohort analysis.

Why Cohort Analysis Matters in Advertising

Advertising platforms are excellent at showing what happened at the moment of conversion. They can tell you which campaign drove a purchase, what the CPA was, and what the immediate return looked like.

But businesses do not grow only on first purchases. They grow on customer quality and customer value over time.

Two campaigns can produce very similar front-end metrics and still be completely different in business value.

For example:

• Campaign A drives a strong first purchase rate but poor repeat purchase behavior
• Campaign B drives a slightly weaker first purchase rate but much stronger repeat revenue

If you only look at immediate ROAS, both campaigns may look similar, or Campaign A may even look better. But if you look at cohorts, you may discover that Campaign B is actually bringing in far more valuable customers.

That is why cohort analysis matters. It reveals the quality of acquisition, not just the quantity of conversions.

The Example

Let’s use a simple e-commerce example.

Every week, the business acquires new customers through paid media. Some of those customers make repeat purchases in the following weeks. Others do not. To understand the difference, the business groups customers by the week of their first purchase and then tracks how much revenue each group generates in the weeks after acquisition.

That produces a table like this:

Cohort (Week of first purchase)	Week 0	Week 1	Week 2	Week 3
Week 1	20	8	5	3
Week 2	22	6	3	1
Week 3	18	10	7	4

 

Assume these values represent revenue per customer in euros.

What the Columns Mean

Before reading the table, the first thing to understand is the column structure.

• Week 0 means the same week the customer was acquired or made their first purchase
• Week 1 means one week after acquisition
• Week 2 means two weeks after acquisition
• Week 3 means three weeks after acquisition

This is critical because these are not shared calendar weeks. They are time offsets from the starting point of each cohort.

That means:

• For the Week 1 cohort, Week 0 is their first week
• For the Week 2 cohort, Week 0 is also their first week
• For the Week 3 cohort, Week 0 is also their first week

So even though these customers entered at different calendar dates, the table lines them up by lifecycle stage.

This is what makes cohort analysis useful. It creates an apples-to-apples comparison.

You are no longer comparing random customers at random moments. You are comparing different customer groups at the exact same point in their relationship with the business.

How to Read the Table Horizontally

The easiest way to begin is to read across one row from left to right.

Take the first row:

• Week 1 cohort, Week 0 = 20
• Week 1 cohort, Week 1 = 8
• Week 1 cohort, Week 2 = 5
• Week 1 cohort, Week 3 = 3

This means customers who first purchased in Week 1 generated:

• €20 per customer in their first week
• €8 per customer in the following week
• €5 per customer in the second week after that
• €3 per customer in the third week after that

So a row tells the story of one cohort over time.

This horizontal view answers a retention and value question:

How does this group behave after acquisition?

When you read rows, you are studying the lifecycle of a specific cohort.

For example, the Week 2 row looks like this:

• 22
• 6
• 3
• 1

That suggests a strong first purchase but weak repeat purchase behavior. The cohort converted well initially, but its value faded quickly.

The Week 3 row looks like this:

• 18
• 10
• 7
• 4

That suggests a lower first purchase than Week 2, but much stronger repeat behavior afterward. This group looks healthier and more valuable over time.

How to Read the Table Vertically

Now move from rows to columns.

This is the part many people miss, but it is where cohort analysis becomes strategically powerful.

Take the Week 1 column:

• Week 1 cohort = 8
• Week 2 cohort = 6
• Week 3 cohort = 10

These numbers do not represent the same calendar week. They represent the same lifecycle moment.

Every number in this column tells you how much customers spent exactly one week after their first purchase.

That is why this is an apples-to-apples comparison.

You are comparing different cohorts at the same stage of their lifecycle.

This vertical view answers a quality question:

Are the customers we are acquiring now better or worse than the customers we acquired earlier?

In this example:

• Customers acquired in Week 2 spent less in their second week than customers acquired in Week 1
• Customers acquired in Week 3 spent more in their second week than both earlier cohorts

That tells you something changed.

Maybe Week 2 brought in lower-quality customers.
Maybe Week 3 brought in better-qualified customers.
Maybe a targeting change, creative update, offer shift, landing page improvement, or CRM flow enhancement improved the quality of acquisition.

This is why the vertical view is so valuable. It helps you understand whether your business is getting better or worse at attracting and keeping the right customers.

The Vertical Quality Check

A useful way to think about columns is this:

Rows tell you the story of one customer group.
Columns tell you the story of how your acquisition quality is evolving.

If you look down a column and later cohorts are performing better, it often means your marketing, product experience, offer, or retention system is improving.

If you look down a column and later cohorts are performing worse, it can be a warning sign that something has declined.

For example, imagine this sequence:

• Week 1 column values go from 8 to 6 to 10

This suggests:

• Week 2 customers were weaker one week after acquisition
• Week 3 customers were stronger one week after acquisition

That pattern invites investigation.

Questions you would ask include:

• Did we change ad creative before Week 3?
• Did we improve the landing page?
• Did we launch a better onboarding email flow?
• Did Week 2 rely too heavily on discount-driven traffic?
• Did Week 3 attract more loyal customers rather than bargain hunters?

The point is that columns do not just compare cohorts. They show whether the business is learning how to bring in better customers over time.

What Week 0 Tells You

Week 0 is especially important because it is your first conversion moment.

For many marketers, Week 0 will feel familiar because it resembles the standard performance view: initial revenue, first purchase value, and front-end return.

In that sense, Week 0 is close to your usual acquisition reporting.

But the real advantage of cohort analysis begins in the columns to the right of Week 0.

That is where you see what immediate reporting misses:

• repeat purchase behavior
• post-acquisition value
• customer stickiness
• long-term profitability

Week 0 tells you whether you converted the customer.
Week 1 and beyond tell you whether you acquired a good customer.

Comparing Total Cohort Value

You can also add across each row to compare total revenue generated by each cohort across the measured period.

Using the table above:

• Week 1 cohort total = 20 + 8 + 5 + 3 = 36
• Week 2 cohort total = 22 + 6 + 3 + 1 = 32
• Week 3 cohort total = 18 + 10 + 7 + 4 = 39

This gives a clear ranking:

• Week 3 cohort is the strongest overall
• Week 1 cohort is in the middle
• Week 2 cohort is the weakest

This matters because the cohort with the highest Week 0 is not always the cohort with the highest total value.

That is one of the most important lessons in performance marketing.

High front-end performance does not always mean high customer quality.

What Business Changes Cohort Analysis Can Reveal

Cohort trends often reflect changes happening across the business, not just inside ad platforms.

Improvements in later cohorts may come from:

• better audience targeting
• stronger creative messaging
• improved landing pages
• more relevant offers
• faster checkout experience
• better email and SMS onboarding
• stronger post-purchase communication
• improved product-market fit

Declines in later cohorts may reflect the opposite:

• lower-quality traffic
• misleading messaging
• overuse of discounts
• weak onboarding
• poor product experience
• fulfillment or site issues

That is why cohort analysis sits at the intersection of marketing, product, retention, and revenue strategy.

How It Helps Performance Marketers Make Better Decisions

Cohort analysis improves decision-making in several ways.

It improves budget allocation

Instead of allocating spend only toward campaigns with the strongest immediate ROAS, you can allocate more confidently toward the campaigns that bring in customers with stronger downstream value.

It improves creative evaluation

A creative that produces cheap conversions but weak repeat behavior may not be as good as it looks. A creative that produces slightly more expensive conversions but stronger repeat revenue may be the smarter long-term winner.

It improves CAC interpretation

A lower CAC is not always better. Sometimes low-cost acquisition brings low-quality customers. Cohort analysis helps reveal whether efficiency gains are coming at the expense of customer value.

It improves retention analysis

If newer cohorts improve in Week 1, Week 2, and Week 3, that may indicate your retention systems are getting stronger. If they worsen, the business may be leaking value after acquisition.

It improves growth quality

Cohort analysis helps distinguish between growth in volume and growth in quality. That distinction matters because not all growth is profitable.

 

 

 

 

 

A Simple Memory Framework

If you want one quick way to remember how to read a cohort table, use this:

Direction	What it shows	The key question
Horizontal	Retention and lifecycle	How does one customer group behave over time?
Vertical	Acquisition quality and business improvement	Are newer customers better or worse than older ones at the same lifecycle stage?

 

That alone will help most marketers read cohort tables much more effectively.

Final Thought

Cohort analysis changes the question from:

How did the campaign perform?

to:

What kind of customers did the campaign bring, and what did they do afterward?

That shift is what makes cohort analysis so valuable.

It helps you move beyond immediate conversion metrics and toward a deeper understanding of customer quality, retention, lifetime value, and true business impact.

In a world where short-term performance metrics can look strong while long-term value quietly declines, cohort analysis gives marketers a clearer and more honest view of what growth actually looks like.

Top of Form

Bottom of Form