The Full-Funnel Google Ads Architecture for DTC

Most brands hit a ceiling at $50K/month. They don't know why.

They're not doing anything wrong. They're just missing the architecture. (And it's rarely explained clearly.)

We audited a DTC brand spending $38K/month on Google Ads. Twelve months later, they were at $250K/month with a 115% revenue lift - without increasing their target ROAS.

The difference wasn't a secret tactic or a new campaign type. It was architecture.

Look, a pattern we see often: brands running Google Ads like separate campaigns fighting for the same budget. They launch Shopping, add some Search, maybe test PMax, and wonder why nothing compounds. We estimate 70% of accounts we audit waste 20-40% of their budget on campaigns that cannibalize each other. (Yes, cannibalize. Their own campaigns are competing with each other.)

What they're missing is the system.

There's a specific architecture that turns fragmented campaigns into a machine where each layer feeds the next. Where Demand Gen builds the audience that Search captures. Where Search validates intent that Shopping converts. Where every dollar spent in one place makes the others work harder.

After managing $184M+ in Google Ads spend across 400+ brands, we've refined this into a framework.

Here's the full breakdown.

The Problem With How Most Brands Run Google Ads

If you're running Search and Shopping separately and seeing results, you're doing what works for most brands at your stage. That's not the problem.

But we noticed something when we zoomed out across 400+ accounts: the brands that scaled past $100K/month were doing something different.

Here's what we see in 90% of accounts we audit - and yes, I know this list will feel familiar:

Shopping campaigns running without feed optimization (the biggest one)
Search campaigns competing with PMax for the same keywords
No Demand Gen or YouTube, so they only capture buyers already searching
Branded campaigns eating budget that should go to prospecting
No clear structure for what each campaign type is supposed to do

The result? Campaigns cannibalize each other. Budget gets wasted on the wrong audiences. And when it's time to scale, nothing works. You push more budget in and ROAS drops. (Sound familiar?)

The core issue is thinking about Google Ads as individual campaigns instead of an integrated system.

When you run Shopping in isolation, you're only capturing people who already know what they want. You show up at the bottom of the funnel and compete on price with everyone else.

When you run Search without proper keyword staging, you're bidding on the same terms as PMax and paying twice for the same clicks.

When you skip Demand Gen entirely, you're invisible until the moment someone searches. By then, your competitors have been building awareness for weeks.

This is why brands commonly hit a ceiling between $20K-50K/month in ad spend and can't break through. The architecture isn't built to compound.

But here's where it gets interesting.

The Compound Stack: A 4-Layer Architecture

The operators running 8-figure brands don't think in campaigns. They think in systems. This framework is how they think.

The system that scales has four distinct layers, each with a specific job:

Layer 1: Demand Gen (Awareness) Job: Reach people before they search. Build interest. Channels: YouTube, YouTube Shorts, Discover, Gmail feeds Budget: 3-5% of total spend (grows as you scale)

Layer 2: Search (Intent Capture) Job: Capture people actively looking for solutions. Qualify intent. Channels: Search campaigns, keyword staging Budget: 20-25% of total spend

Layer 3: Shopping/PMax (Conversion) Job: Convert high-intent buyers. Maximize ROAS on ready traffic. Channels: Performance Max, Standard Shopping, feed-driven campaigns Budget: 65-75% of total spend

Layer 4: Retargeting (Recovery) Job: Re-engage visitors, recover abandoned carts, drive repeat purchases. Channels: Demand Gen retargeting, Display Budget: 3-5% of total spend

Here's how this maps to a real account.

One of our clients spending $500K/month runs this structure:

85% prospecting (Layer 2 + Layer 3)
10% branded (protecting their name)
3% retargeting (Layer 4)

Within prospecting, the breakdown is:

70% PMax (3 campaigns for top products + 1 catch-all)
25% Search (2 branded + intent keywords)
5% Demand Gen (awareness + retargeting per country)
2% Standard Shopping (specific product tests)

This isn't arbitrary. It's calibrated based on where the highest-leverage conversions happen at their scale.

Layer 1 - Demand Gen (The Pre-Search Play)

The default approach is to show up when customers are ready to buy.

By then, you've already missed the window.

Demand Gen exists to reach people before they search. When they're browsing YouTube, scrolling Discover, checking Gmail. When they're in consideration mode but haven't typed a keyword yet.

Here's why this matters:

YouTube Shorts alone has 650M+ monthly viewers. 70% say Shorts content influences their purchase decisions - 45 points higher than competing platforms.

More importantly, 45% of the audience you reach on YouTube Shorts is unique. You can't reach them on Instagram Reels.

If you're only running Search and Shopping, you're invisible to nearly half your potential customers until the moment they search. And by then, you're competing with everyone else who showed up late.

How to run Demand Gen effectively:

Start with video. YouTube is the core placement. Create 15-60 second content that tells a story about the problem you solve. Not product demos. Stories.

Target lookalikes of your best customers. Use your first-party data (email lists, past purchasers) to build audiences that match your highest-LTV buyers.

Measure the right things. Demand Gen looks terrible in last-click attribution. We'll cover this in the attribution section. For now, understand that you're measuring influence, not direct conversions.

Budget allocation: Start at 3-5% of total spend. This isn't your conversion engine. It's your awareness engine that makes your conversion campaigns work harder.

The synergy:

A client running Demand Gen alongside Search and Shopping saw their Brand Search ROAS hit 10x - because Demand Gen was warming audiences that later converted via branded queries.

Without Demand Gen, they'd only see the branded search conversion and think that's where the value was created. It wasn't. The value was created weeks earlier when someone saw their YouTube ad.

If you've ever wondered why your branded search ROAS keeps climbing while your prospecting stays flat, this is probably why.

Demand Gen works. But only if you measure it correctly. (And the default measurement approach misses it entirely.)

Layer 2 - Search (Intent Staging)

Search is where you capture demand that already exists.

But there are common patterns that limit its effectiveness.

They bid on broad keywords without staging. They compete with their own PMax campaigns. They skip intent-specific landing pages.

Here's the framework that works:

The Intent Ladder:

Not all search intent is equal. Stage your campaigns by intent level:

Stage 1: Competitor-Aware Keywords: [competitor name] alternative, [competitor] vs, better than [competitor] These people know the category. They're shopping options. Highest intent after brand searches.

Stage 2: Solution-Aware Keywords: best [product type], [product type] for [use case], top [product type] 2026 They know what they want. They don't know who to buy from.

Stage 3: Problem-Aware Keywords: how to [solve problem], why does [problem] happen Early stage. They know the pain, not the solution.

Stage 4: Unaware Usually not Search-appropriate. This is Demand Gen territory.

Run separate campaigns for each stage. Different bids, different ads, different landing pages.

A DTC furniture brand we worked with was stuck at 1.3% conversion rate across all Search. We rebuilt with intent-specific landers - matching the landing page messaging to the exact search intent.

Five weeks later: 2.8% conversion rate. 115% revenue lift. Same budget. (The client thought we'd changed their bidding strategy. We'd changed their landing pages. That's it.)

The threshold:

Don't start with Search if you're below $10K/month total spend. You need enough volume for the algorithm to learn. Below that threshold, focus on Shopping/PMax where purchase intent is clearest.

Above $60K/month, Search becomes essential for capturing the demand your other campaigns create.

If you're nodding because you've seen campaigns that work alone but underperform together, you already know Search without staging is leaking money.

Layer 3 - Shopping and PMax (The Conversion Engine)

This is where 65-75% of your budget lives.

PMax and Shopping are your conversion workhorses. They capture high-intent buyers who know what they want and are ready to purchase.

But here's what often gets overlooked:

80% of Shopping performance comes from your feed.

Not your bids. Not your audiences. Your feed.

We've seen accounts drop CPCs from $3+ to $0.80 just by optimizing product titles and attributes. (It's unglamorous work. Nobody tweets about product feed optimization. But it drives more impact than almost anything else.)

Feed optimization priorities (in order):

Titles: Format as [Brand] [Product Type] [Key Attribute]. Search-optimized, not your internal naming conventions.
Product types and categories: Google's taxonomy, properly mapped. Don't skip this.
Custom labels: Segment by margin, bestsellers, seasonality. This lets you bid differently on high-margin vs. low-margin products.
GTINs: If you have them, use them. Google trusts products with valid identifiers.
Image quality: Primary image should be clean, white background, product clearly visible. This isn't creative - it's feed hygiene.

PMax structure for scale:

At $500K/month, our client runs:

3 PMax campaigns for top-performing product categories
1 catch-all PMax for the long tail
2 Search campaigns (branded + non-branded)
1 Standard Shopping campaign for specific product tests

The key is segmentation. Don't dump everything into one PMax campaign. Segment by:

Product category (different ROAS targets)
Margin (high-margin products can afford higher bids)
Performance (proven winners vs. tests)

Avoiding PMax conflicts:

PMax will cannibalize your branded search if you let it. It'll steal credit for conversions it didn't earn.

Solutions:

Exclude brand terms from PMax (negative keywords at account level)
Run branded Search separately with higher priority
Set PMax tROAS 5x higher than actual target to force selectivity

The staging threshold:

Shopping and PMax should be your starting point if you're below $60K/month. Build a stable base here before adding Search and Demand Gen layers.

Above $60K/month, start layering. The ecosystem compounds.

At this point, the pattern is clear: isolated campaigns = isolated results. The structure works - until you try to measure it with standard tools.

Layer 4 - Retargeting (Recovery and Repeat)

Retargeting gets 3-5% of budget. Not more.

Here's why: if you're spending heavily on retargeting, your prospecting is broken.

Retargeting exists to recover people who almost converted and to re-engage past customers. It's not a primary growth lever.

What to run:

Abandoned cart sequences (7-14 day window)
Past purchaser re-engagement (30-60 days post-purchase)
High-intent visitors who viewed 3+ products

What to avoid:

Retargeting everyone who visited any page
Long retargeting windows (30+ days for non-purchasers)
Spending more than 5% here

If retargeting is driving most of your conversions, that means your cold traffic campaigns aren't working. Fix prospecting first.

The Attribution Reality

Here's where measurement trips up even experienced operators. (And we say this from experience - we got it wrong for longer than we'd like to admit.)

Your YouTube and Demand Gen campaigns will show terrible ROAS in platform dashboards. Triple Whale, GA4, in-app reporting - they'll all say these campaigns are losing money.

They're lying.

Not intentionally. But last-click attribution fundamentally cannot measure top-funnel impact. It's like judging a basketball team only by who makes the final shot - you miss everything that created the opportunity.

The actual data:

YouTube ads show "terrible ROAS" in dashboards. But when we measure correctly:

Branded search volume up 40%
Cold Shopping campaigns converting at 2x higher rate
10x ROAS on Brand Search from YouTube-influenced traffic

The YouTube campaign created this value. But last-click attribution gives all the credit to the branded search that came later. (This is why so many brands kill their YouTube campaigns and then wonder why everything else drops.)

As one practitioner put it: "Track branded search volume to capture half of the ad performance your dashboards are missing."

What to track instead:

Branded search volume correlation: Run YouTube for 4-6 weeks. Measure branded search volume before, during, and after. The lift tells you what last-click can't.
Holdout tests: Run campaigns in specific geos or cohorts. Compare conversion rates for exposed vs. unexposed audiences. This is the gold standard.
Cross-platform tracking: Use custom conversions to track Google traffic post-Meta ads (and vice versa). UTM for Brand Search from Meta lift.
Marketing Mix Modeling (MMM): Demand for MMM surged 300% in 2025. It's no longer just for enterprise. Granular MMM now works at campaign level for DTC.

The mindset shift:

Stop optimizing each campaign for direct ROAS.

Start optimizing the system for total revenue.

A campaign that looks like a 1.5x ROAS loser might be driving 40% of your branded search conversions. Kill it and watch your "efficient" campaigns crater.

Sequencing - When to Add Each Layer

Don't try to build everything at once. Layer it.

Stage 1: Foundation ($0-$30K/month)

Start with Shopping/PMax only.

Focus on:

Feed optimization (this is 80% of your results)
Product segmentation
Understanding what converts

Don't add: Search, Demand Gen, YouTube

Why: You need conversion data before you can optimize higher-funnel campaigns. Build a stable base first.

Success metric: Profitable ROAS at current spend level.

Stage 2: Intent Capture ($30K-$100K/month)

Add Search campaigns.

Structure:

Branded search (protect your name)
Non-branded intent keywords (competitor-aware, solution-aware)
Intent-specific landing pages

Don't add yet: Demand Gen, YouTube

Why: At this stage, you're capturing existing demand. Search finds people already looking. Demand Gen creates demand you can't yet absorb.

Success metric: 50+ conversions per month per campaign. This is the threshold for reliable optimization.

Stage 3: Demand Creation ($100K+/month)

Add Demand Gen and YouTube.

Structure:

YouTube video campaigns (awareness)
Demand Gen for prospecting (lookalikes)
Demand Gen for retargeting (per-country if international)

Why now: You have enough conversion volume to measure influence. You have stable bottom-funnel campaigns to capture the demand you create.

Success metric: Branded search volume lift, overall ROAS improvement (not campaign-level).

We learned this the hard way. One client launched YouTube at $30K/month spend before their Shopping campaigns were stable. No capture campaigns ready to absorb the demand. $47K in awareness spend over 6 weeks, minimal conversions. The traffic was there. The system to absorb it wasn't.

Now we don't add Demand Gen until the conversion foundation is profitable. Expensive lesson, but it shaped everything we do.

The T.K.S. Method:

For each new campaign type:

Test: Split budget into 5-10 test variations. Start lean.

Kill: At CAC threshold (usually 1.5-2x target), kill underperformers. No emotional attachment.

Scale: Winners get budget. Losers get cut. No middle ground.

One brand used this to turn $3.7K test spend into 1,463 purchases. They tested 10 campaign structures, killed 7, and scaled 3.

Common Mistakes and How to Avoid Them

After auditing hundreds of accounts, these are the patterns we see repeatedly:

Mistake 1: Starting with Demand Gen too early

Brands hear "full-funnel" and immediately add YouTube. But if your Shopping campaigns aren't profitable, you don't need more traffic. You need better conversion rates.

Fix: Don't add Demand Gen until you're spending $60K+/month profitably.

Mistake 2: Running PMax without exclusions

PMax will happily spend your budget on branded searches and take credit for conversions it didn't earn. Then you'll think PMax is amazing while your actual prospecting suffers.

Fix: Exclude brand terms. Run branded Search separately. Set PMax tROAS artificially high (5x target) to force it into true prospecting.

Mistake 3: Ignoring feed optimization

We see brands spending $100K/month on Shopping with product titles like "SKU-4829-BLK-LG".

Feed optimization drives 80% of Shopping results. Titles should be [Brand] [Product Type] [Key Attribute]. Descriptions should include search terms. Custom labels should segment by margin.

Fix: Before touching bids or audiences, fix your feed. CPCs drop. Impressions increase. ROAS improves.

Mistake 4: Measuring Demand Gen by direct ROAS

If you judge YouTube by last-click ROAS, you'll kill it within a week. It looks like a money pit.

Measure branded search lift instead. Measure holdout tests. Measure overall revenue, not campaign-level returns.

Fix: Give Demand Gen 6-8 weeks before judging. Measure influence, not attribution.

Mistake 5: Over-investing in retargeting

If retargeting is driving more than 10% of conversions, your cold traffic is broken.

Retargeting is recovery, not growth. 3-5% of budget, max.

Fix: Cut retargeting to 5%. Reinvest in prospecting. Fix the top of funnel.

Implementation Checklist

Here's the sequence to implement this architecture:

Week 1-2: Foundation Audit

Audit current feed (titles, categories, GTINs, images)
Implement feed optimizations
Structure Shopping/PMax campaigns by category and margin
Set up brand exclusions in PMax
Establish baseline branded search volume

Week 3-4: Search Layer

Launch branded Search campaign
Identify top 20 non-branded intent keywords
Create intent-specific landing pages
Launch non-branded Search campaigns by intent stage
Set up keyword exclusions between campaigns

Week 5-8: Measurement Foundation

Set up branded search volume tracking
Create custom conversions for cross-platform tracking
Establish holdout methodology for future tests
Document baseline metrics for all campaigns

Week 8+: Demand Gen Layer (if at scale)

Create 3-5 video assets (15-60 seconds)
Build lookalike audiences from best customers
Launch YouTube campaigns
Launch Demand Gen prospecting campaigns
Measure branded search lift weekly

The brand we audited at $38K? They're at $250K now. Same team. Same products. Different architecture.

The difference between brands stuck at $50K/month and brands scaling to $500K/month isn't usually the tactics. It's the architecture.

Isolated campaigns fight each other. Integrated systems compound. (This is the whole insight, really.)

In 12 months, you'll be somewhere. The question is whether you built the system that compounds - or kept running campaigns that compete with themselves.

Brands that stay at $50K today and don't change their structure will likely be there in 12 months. Not because they're doing something wrong. Because the architecture that enables the breakthrough isn't in place yet.

The layers exist. The sequence is clear. The measurement framework works. The only variable is when you start.

Build the foundation. Layer the campaigns. Measure the system, not the silos.

That's the whole game.

Run this on your account

Google Ads agent · 6 minutes.

We'll run this exact diagnostic on your live PMax and send back the results. No call required.

Run agent →

Ruslan Galba

Google Ads + AI · Tegra

Ruslan co-founded Tegra in 2017. Runs the Google Ads practice - feed, PMax, search, attribution. Writes weekly about the parts of paid search operators are afraid to touch.

All articles →

Most brands hit a ceiling at $50K/month. They don't know why.

They're not doing anything wrong. They're just missing the architecture. (And it's rarely explained clearly.)

We audited a DTC brand spending $38K/month on Google Ads. Twelve months later, they were at $250K/month with a 115% revenue lift - without increasing their target ROAS.

The difference wasn't a secret tactic or a new campaign type. It was architecture.

What they're missing is the system.

After managing $184M+ in Google Ads spend across 400+ brands, we've refined this into a framework.

Here's the full breakdown.

The Problem With How Most Brands Run Google Ads

If you're running Search and Shopping separately and seeing results, you're doing what works for most brands at your stage. That's not the problem.

But we noticed something when we zoomed out across 400+ accounts: the brands that scaled past $100K/month were doing something different.

Here's what we see in 90% of accounts we audit - and yes, I know this list will feel familiar:

Shopping campaigns running without feed optimization (the biggest one)
Search campaigns competing with PMax for the same keywords
No Demand Gen or YouTube, so they only capture buyers already searching
Branded campaigns eating budget that should go to prospecting
No clear structure for what each campaign type is supposed to do

The result? Campaigns cannibalize each other. Budget gets wasted on the wrong audiences. And when it's time to scale, nothing works. You push more budget in and ROAS drops. (Sound familiar?)

The core issue is thinking about Google Ads as individual campaigns instead of an integrated system.

When you run Shopping in isolation, you're only capturing people who already know what they want. You show up at the bottom of the funnel and compete on price with everyone else.

When you run Search without proper keyword staging, you're bidding on the same terms as PMax and paying twice for the same clicks.

When you skip Demand Gen entirely, you're invisible until the moment someone searches. By then, your competitors have been building awareness for weeks.

This is why brands commonly hit a ceiling between $20K-50K/month in ad spend and can't break through. The architecture isn't built to compound.

But here's where it gets interesting.

The Compound Stack: A 4-Layer Architecture

The operators running 8-figure brands don't think in campaigns. They think in systems. This framework is how they think.

The system that scales has four distinct layers, each with a specific job:

Layer 1: Demand Gen (Awareness) Job: Reach people before they search. Build interest. Channels: YouTube, YouTube Shorts, Discover, Gmail feeds Budget: 3-5% of total spend (grows as you scale)

Layer 2: Search (Intent Capture) Job: Capture people actively looking for solutions. Qualify intent. Channels: Search campaigns, keyword staging Budget: 20-25% of total spend

Layer 4: Retargeting (Recovery) Job: Re-engage visitors, recover abandoned carts, drive repeat purchases. Channels: Demand Gen retargeting, Display Budget: 3-5% of total spend

Here's how this maps to a real account.

One of our clients spending $500K/month runs this structure:

85% prospecting (Layer 2 + Layer 3)
10% branded (protecting their name)
3% retargeting (Layer 4)

Within prospecting, the breakdown is:

70% PMax (3 campaigns for top products + 1 catch-all)
25% Search (2 branded + intent keywords)
5% Demand Gen (awareness + retargeting per country)
2% Standard Shopping (specific product tests)

This isn't arbitrary. It's calibrated based on where the highest-leverage conversions happen at their scale.

Layer 1 - Demand Gen (The Pre-Search Play)

The default approach is to show up when customers are ready to buy.

By then, you've already missed the window.

Demand Gen exists to reach people before they search. When they're browsing YouTube, scrolling Discover, checking Gmail. When they're in consideration mode but haven't typed a keyword yet.

Here's why this matters:

YouTube Shorts alone has 650M+ monthly viewers. 70% say Shorts content influences their purchase decisions - 45 points higher than competing platforms.

More importantly, 45% of the audience you reach on YouTube Shorts is unique. You can't reach them on Instagram Reels.

If you're only running Search and Shopping, you're invisible to nearly half your potential customers until the moment they search. And by then, you're competing with everyone else who showed up late.

How to run Demand Gen effectively:

Start with video. YouTube is the core placement. Create 15-60 second content that tells a story about the problem you solve. Not product demos. Stories.

Target lookalikes of your best customers. Use your first-party data (email lists, past purchasers) to build audiences that match your highest-LTV buyers.

Budget allocation: Start at 3-5% of total spend. This isn't your conversion engine. It's your awareness engine that makes your conversion campaigns work harder.

The synergy:

A client running Demand Gen alongside Search and Shopping saw their Brand Search ROAS hit 10x - because Demand Gen was warming audiences that later converted via branded queries.

Without Demand Gen, they'd only see the branded search conversion and think that's where the value was created. It wasn't. The value was created weeks earlier when someone saw their YouTube ad.

If you've ever wondered why your branded search ROAS keeps climbing while your prospecting stays flat, this is probably why.

Demand Gen works. But only if you measure it correctly. (And the default measurement approach misses it entirely.)

Layer 2 - Search (Intent Staging)

Search is where you capture demand that already exists.

But there are common patterns that limit its effectiveness.

They bid on broad keywords without staging. They compete with their own PMax campaigns. They skip intent-specific landing pages.

Here's the framework that works:

The Intent Ladder:

Not all search intent is equal. Stage your campaigns by intent level:

Stage 2: Solution-Aware Keywords: best [product type], [product type] for [use case], top [product type] 2026 They know what they want. They don't know who to buy from.

Stage 3: Problem-Aware Keywords: how to [solve problem], why does [problem] happen Early stage. They know the pain, not the solution.

Stage 4: Unaware Usually not Search-appropriate. This is Demand Gen territory.

Run separate campaigns for each stage. Different bids, different ads, different landing pages.

A DTC furniture brand we worked with was stuck at 1.3% conversion rate across all Search. We rebuilt with intent-specific landers - matching the landing page messaging to the exact search intent.

Five weeks later: 2.8% conversion rate. 115% revenue lift. Same budget. (The client thought we'd changed their bidding strategy. We'd changed their landing pages. That's it.)

The threshold:

Don't start with Search if you're below $10K/month total spend. You need enough volume for the algorithm to learn. Below that threshold, focus on Shopping/PMax where purchase intent is clearest.

Above $60K/month, Search becomes essential for capturing the demand your other campaigns create.

If you're nodding because you've seen campaigns that work alone but underperform together, you already know Search without staging is leaking money.

Layer 3 - Shopping and PMax (The Conversion Engine)

This is where 65-75% of your budget lives.

PMax and Shopping are your conversion workhorses. They capture high-intent buyers who know what they want and are ready to purchase.

But here's what often gets overlooked:

80% of Shopping performance comes from your feed.

Not your bids. Not your audiences. Your feed.

Feed optimization priorities (in order):

Titles: Format as [Brand] [Product Type] [Key Attribute]. Search-optimized, not your internal naming conventions.
Product types and categories: Google's taxonomy, properly mapped. Don't skip this.
Custom labels: Segment by margin, bestsellers, seasonality. This lets you bid differently on high-margin vs. low-margin products.
GTINs: If you have them, use them. Google trusts products with valid identifiers.
Image quality: Primary image should be clean, white background, product clearly visible. This isn't creative - it's feed hygiene.

PMax structure for scale:

At $500K/month, our client runs:

3 PMax campaigns for top-performing product categories
1 catch-all PMax for the long tail
2 Search campaigns (branded + non-branded)
1 Standard Shopping campaign for specific product tests

The key is segmentation. Don't dump everything into one PMax campaign. Segment by:

Product category (different ROAS targets)
Margin (high-margin products can afford higher bids)
Performance (proven winners vs. tests)

Avoiding PMax conflicts:

PMax will cannibalize your branded search if you let it. It'll steal credit for conversions it didn't earn.

Solutions:

Exclude brand terms from PMax (negative keywords at account level)
Run branded Search separately with higher priority
Set PMax tROAS 5x higher than actual target to force selectivity

The staging threshold:

Shopping and PMax should be your starting point if you're below $60K/month. Build a stable base here before adding Search and Demand Gen layers.

Above $60K/month, start layering. The ecosystem compounds.

At this point, the pattern is clear: isolated campaigns = isolated results. The structure works - until you try to measure it with standard tools.

Layer 4 - Retargeting (Recovery and Repeat)

Retargeting gets 3-5% of budget. Not more.

Here's why: if you're spending heavily on retargeting, your prospecting is broken.

Retargeting exists to recover people who almost converted and to re-engage past customers. It's not a primary growth lever.

What to run:

Abandoned cart sequences (7-14 day window)
Past purchaser re-engagement (30-60 days post-purchase)
High-intent visitors who viewed 3+ products

What to avoid:

Retargeting everyone who visited any page
Long retargeting windows (30+ days for non-purchasers)
Spending more than 5% here

If retargeting is driving most of your conversions, that means your cold traffic campaigns aren't working. Fix prospecting first.

The Attribution Reality

Here's where measurement trips up even experienced operators. (And we say this from experience - we got it wrong for longer than we'd like to admit.)

Your YouTube and Demand Gen campaigns will show terrible ROAS in platform dashboards. Triple Whale, GA4, in-app reporting - they'll all say these campaigns are losing money.

They're lying.

The actual data:

YouTube ads show "terrible ROAS" in dashboards. But when we measure correctly:

Branded search volume up 40%
Cold Shopping campaigns converting at 2x higher rate
10x ROAS on Brand Search from YouTube-influenced traffic

As one practitioner put it: "Track branded search volume to capture half of the ad performance your dashboards are missing."

What to track instead:

Branded search volume correlation: Run YouTube for 4-6 weeks. Measure branded search volume before, during, and after. The lift tells you what last-click can't.
Holdout tests: Run campaigns in specific geos or cohorts. Compare conversion rates for exposed vs. unexposed audiences. This is the gold standard.
Cross-platform tracking: Use custom conversions to track Google traffic post-Meta ads (and vice versa). UTM for Brand Search from Meta lift.
Marketing Mix Modeling (MMM): Demand for MMM surged 300% in 2025. It's no longer just for enterprise. Granular MMM now works at campaign level for DTC.

The mindset shift:

Stop optimizing each campaign for direct ROAS.

Start optimizing the system for total revenue.

A campaign that looks like a 1.5x ROAS loser might be driving 40% of your branded search conversions. Kill it and watch your "efficient" campaigns crater.

Sequencing - When to Add Each Layer

Don't try to build everything at once. Layer it.

Stage 1: Foundation ($0-$30K/month)

Start with Shopping/PMax only.

Focus on:

Feed optimization (this is 80% of your results)
Product segmentation
Understanding what converts

Don't add: Search, Demand Gen, YouTube

Why: You need conversion data before you can optimize higher-funnel campaigns. Build a stable base first.

Success metric: Profitable ROAS at current spend level.

Stage 2: Intent Capture ($30K-$100K/month)

Add Search campaigns.

Structure:

Branded search (protect your name)
Non-branded intent keywords (competitor-aware, solution-aware)
Intent-specific landing pages

Don't add yet: Demand Gen, YouTube

Why: At this stage, you're capturing existing demand. Search finds people already looking. Demand Gen creates demand you can't yet absorb.

Success metric: 50+ conversions per month per campaign. This is the threshold for reliable optimization.

Stage 3: Demand Creation ($100K+/month)

Add Demand Gen and YouTube.

Structure:

YouTube video campaigns (awareness)
Demand Gen for prospecting (lookalikes)
Demand Gen for retargeting (per-country if international)

Why now: You have enough conversion volume to measure influence. You have stable bottom-funnel campaigns to capture the demand you create.

Success metric: Branded search volume lift, overall ROAS improvement (not campaign-level).

Now we don't add Demand Gen until the conversion foundation is profitable. Expensive lesson, but it shaped everything we do.

The T.K.S. Method:

For each new campaign type:

Test: Split budget into 5-10 test variations. Start lean.

Kill: At CAC threshold (usually 1.5-2x target), kill underperformers. No emotional attachment.

Scale: Winners get budget. Losers get cut. No middle ground.

One brand used this to turn $3.7K test spend into 1,463 purchases. They tested 10 campaign structures, killed 7, and scaled 3.

Common Mistakes and How to Avoid Them

After auditing hundreds of accounts, these are the patterns we see repeatedly:

Mistake 1: Starting with Demand Gen too early

Brands hear "full-funnel" and immediately add YouTube. But if your Shopping campaigns aren't profitable, you don't need more traffic. You need better conversion rates.

Fix: Don't add Demand Gen until you're spending $60K+/month profitably.

Mistake 2: Running PMax without exclusions

PMax will happily spend your budget on branded searches and take credit for conversions it didn't earn. Then you'll think PMax is amazing while your actual prospecting suffers.

Fix: Exclude brand terms. Run branded Search separately. Set PMax tROAS artificially high (5x target) to force it into true prospecting.

Mistake 3: Ignoring feed optimization

We see brands spending $100K/month on Shopping with product titles like "SKU-4829-BLK-LG".

Feed optimization drives 80% of Shopping results. Titles should be [Brand] [Product Type] [Key Attribute]. Descriptions should include search terms. Custom labels should segment by margin.

Fix: Before touching bids or audiences, fix your feed. CPCs drop. Impressions increase. ROAS improves.

Mistake 4: Measuring Demand Gen by direct ROAS

If you judge YouTube by last-click ROAS, you'll kill it within a week. It looks like a money pit.

Measure branded search lift instead. Measure holdout tests. Measure overall revenue, not campaign-level returns.

Fix: Give Demand Gen 6-8 weeks before judging. Measure influence, not attribution.

Mistake 5: Over-investing in retargeting

If retargeting is driving more than 10% of conversions, your cold traffic is broken.

Retargeting is recovery, not growth. 3-5% of budget, max.

Fix: Cut retargeting to 5%. Reinvest in prospecting. Fix the top of funnel.

Implementation Checklist

Here's the sequence to implement this architecture:

Week 1-2: Foundation Audit

Audit current feed (titles, categories, GTINs, images)
Implement feed optimizations
Structure Shopping/PMax campaigns by category and margin
Set up brand exclusions in PMax
Establish baseline branded search volume

Week 3-4: Search Layer

Launch branded Search campaign
Identify top 20 non-branded intent keywords
Create intent-specific landing pages
Launch non-branded Search campaigns by intent stage
Set up keyword exclusions between campaigns

Week 5-8: Measurement Foundation

Set up branded search volume tracking
Create custom conversions for cross-platform tracking
Establish holdout methodology for future tests
Document baseline metrics for all campaigns

Week 8+: Demand Gen Layer (if at scale)

Create 3-5 video assets (15-60 seconds)
Build lookalike audiences from best customers
Launch YouTube campaigns
Launch Demand Gen prospecting campaigns
Measure branded search lift weekly

The brand we audited at $38K? They're at $250K now. Same team. Same products. Different architecture.

The difference between brands stuck at $50K/month and brands scaling to $500K/month isn't usually the tactics. It's the architecture.

Isolated campaigns fight each other. Integrated systems compound. (This is the whole insight, really.)

In 12 months, you'll be somewhere. The question is whether you built the system that compounds - or kept running campaigns that compete with themselves.

The layers exist. The sequence is clear. The measurement framework works. The only variable is when you start.

Build the foundation. Layer the campaigns. Measure the system, not the silos.

That's the whole game.

Run this on your account

Google Ads agent · 6 minutes.

We'll run this exact diagnostic on your live PMax and send back the results. No call required.

Run agent →

Ruslan Galba

Google Ads + AI · Tegra

Ruslan co-founded Tegra in 2017. Runs the Google Ads practice - feed, PMax, search, attribution. Writes weekly about the parts of paid search operators are afraid to touch.

All articles →

The Full-Funnel Google Ads Architecture for DTC

The Problem With How Most Brands Run Google Ads

The Compound Stack: A 4-Layer Architecture

Layer 1 - Demand Gen (The Pre-Search Play)

Layer 2 - Search (Intent Staging)

Layer 3 - Shopping and PMax (The Conversion Engine)

Layer 4 - Retargeting (Recovery and Repeat)

The Attribution Reality

Sequencing - When to Add Each Layer

Common Mistakes and How to Avoid Them

Implementation Checklist

Google Ads agent · 6 minutes.

Have founders run your accounts.
Or run them yourself.

The Full-Funnel Google Ads Architecture for DTC

The Problem With How Most Brands Run Google Ads

The Compound Stack: A 4-Layer Architecture

Layer 1 - Demand Gen (The Pre-Search Play)

Layer 2 - Search (Intent Staging)

Layer 3 - Shopping and PMax (The Conversion Engine)

Layer 4 - Retargeting (Recovery and Repeat)

The Attribution Reality

Sequencing - When to Add Each Layer

Common Mistakes and How to Avoid Them

Implementation Checklist

Google Ads agent · 6 minutes.

Have founders run your accounts.
Or run them yourself.

The Problem With How Most Brands Run Google Ads

The Compound Stack: A 4-Layer Architecture

Layer 1 - Demand Gen (The Pre-Search Play)

Layer 2 - Search (Intent Staging)

Layer 3 - Shopping and PMax (The Conversion Engine)

Layer 4 - Retargeting (Recovery and Repeat)

The Attribution Reality

Sequencing - When to Add Each Layer

Common Mistakes and How to Avoid Them

Implementation Checklist

Google Ads agent · 6 minutes.

Have founders run your accounts.Or run them yourself.

The Problem With How Most Brands Run Google Ads

The Compound Stack: A 4-Layer Architecture

Layer 1 - Demand Gen (The Pre-Search Play)

Layer 2 - Search (Intent Staging)

Layer 3 - Shopping and PMax (The Conversion Engine)

Layer 4 - Retargeting (Recovery and Repeat)

The Attribution Reality

Sequencing - When to Add Each Layer

Common Mistakes and How to Avoid Them

Implementation Checklist

Google Ads agent · 6 minutes.

Have founders run your accounts.Or run them yourself.

Have founders run your accounts.
Or run them yourself.

Have founders run your accounts.
Or run them yourself.