Media Company Automates Attribution Across 12 Campaigns

Challenge

A Nordic media company runs a subscription business.

12 acquisition campaigns across multiple channels:

• Google Search (brand + non-brand)
• Google Display
• YouTube
• Meta (Facebook + Instagram)
• LinkedIn
• Native advertising
• Podcast sponsorships
• Content partnerships

The problem:

Each channel claims credit for the same conversions.

What their dashboards showed:

• Google Ads: "We drove 450 conversions this month"
• Meta Ads: "We drove 380 conversions this month"
• LinkedIn: "We drove 95 conversions this month"

Total claimed: 925 conversions

Actual conversions: 520

178% over-attribution.

Everyone claiming credit. No one knowing the truth.

The business impact:

Budget decisions were impossible:

• Which channel actually works?
• Where to add budget?
• Where to cut?
• What's the true CAC per channel?

They were optimizing blind.

Increasing spend on channels that looked good in isolation. But maybe only touching customers other channels already warmed up.

Classic attribution nightmare.

The Manual Attribution Attempt

They tried fixing it themselves.

Hired a data analyst ($95K/year).

Month 1-3: Building the Model

Analyst worked on:

• Extracting data from all platforms
• Loading to BigQuery
• Matching user journeys
• Building attribution rules
• Creating dashboards

Challenges:

• Each platform tracks differently
• Cookie tracking incomplete
• Cross-device journeys messy
• Which touchpoint gets credit?

They chose: Time-decay attribution (More recent touchpoints get more credit)

Month 4: First Results

Built Tableau dashboard showing:

• Multi-touch journeys
• Attribution by channel
• "Truth" version of performance

Problems:

• Static view (weekly updates only)
• Complex to understand
• Didn't adapt to changes
• Still unclear how to optimize

Month 5: Reality Check

Campaign manager: "So what should I do with this?"

Analyst: "Well, YouTube gets 12% attribution but claims 15%..."

Campaign manager: "OK... so reduce YouTube by 3%?"

Analyst: "Maybe? It's not that simple..."

The attribution model answered "what happened" But not "what to do next"

Plus:

• Analyst spent full-time maintaining model
• Dashboards broke when campaigns changed
• Team still guessing on optimization
• €95K/year for confusion

Solution

They connected Cogny as complementary tool.

Setup: 15 minutes

• All 12 campaigns connected
• GA4 + BigQuery for complete journey data
• Historical data imported

48 Hours Later:

AI analyzed complete customer journeys.

Built multi-touch attribution automatically.

But more importantly:

Generated actionable tickets based on attribution insights.

Ticket Examples

Ticket #1: YouTube is Undervalued

AI Analysis: "YouTube appears to drive only 40 direct conversions (last-click).

But journey analysis shows YouTube touchpoints appear in 180 conversion paths (35% of total).

Average position: 2.1 (early in journey).

Role: Awareness and consideration.

Budget allocation: 8% (€8,000/month) Suggested allocation: 15% (€15,000/month)

Expected impact: +65 conversions/month"

This is actionable.

Not just "YouTube has 12% attribution." But "Increase YouTube budget to €15K, expect +65 conversions."

Ticket #2: LinkedIn Overlap with Google Search

AI Analysis: "73% of LinkedIn conversions also touched Google Search (brand).

LinkedIn introduces customers who then search brand.

LinkedIn's true value: awareness that drives brand search.

Don't optimize LinkedIn for direct conversions. Optimize for brand search lift.

Recommendation: Track brand search volume as LinkedIn success metric."

This changes strategy.

They were about to pause LinkedIn (expensive, low "direct" ROI). But AI showed it drives valuable brand awareness.

Ticket #3: Meta is Last-Touch Hero

AI Analysis: "Meta gets credit for 380 conversions (last-click).

But 290 of these (76%) had prior touchpoints from other channels.

Meta's role: Conversion (not acquisition).

Meta works best as retargeting, not cold acquisition.

Recommendation: Shift Meta budget from cold to warm audiences."

Ticket #4: Content Partnerships Hidden Value

AI Analysis: "Content partnerships appear in 45% of high-value subscriber journeys (LTV >€500).

But only 8% of low-value journeys (LTV <€200).

Content partnerships attract quality, not quantity.

Current budget: €6,000/month Suggested: €12,000/month (optimize for quality)"

Ticket #5: Podcast Timing

AI Analysis: "Podcast touchpoints → conversion averages 18 days.

Longest delay of all channels.

Podcast builds long-term awareness.

Don't evaluate on immediate conversions. Track brand lift and assisted conversions."

12 more tickets with specific, channel-specific insights.

Implementation

Week 1: Strategy Shift

Realized their attribution model was correct about the numbers. But they didn't know what to DO with the numbers.

AI told them exactly what to do.

Changes made:

YouTube:

• Increased budget from €8K to €15K/month
• Optimized for view-through, not click-through
• New KPI: Brand search lift

LinkedIn:

• Kept budget stable (was going to cut)
• Changed success metric to brand awareness
• New KPI: Brand search volume

Meta:

• Reduced cold acquisition budget 40%
• Increased retargeting budget 60%
• New focus: Convert warm audiences

Content Partnerships:

• Doubled budget (€6K → €12K)
• Focused on quality publishers
• Optimized for high-LTV subscribers

Podcast:

• Changed evaluation period from 7 days to 30 days
• Tracked brand lift instead of direct conversions
• Continued investment with new metrics

Weeks 2-4: See Results

Conversions increased across the board.

Why:

• YouTube driving more awareness
• LinkedIn building brand properly
• Meta converting effectively
• Content partnerships bringing quality
• Podcast building long-term value

All optimized for their actual role.

Not for last-click conversions.

Results

After 3 Months

Attribution Clarity:

• Before: 178% over-attribution (everyone claiming same conversions)
• After: 100% proper attribution (credit distributed correctly)
• Result: Truth

Channel Performance (Properly Attributed):

YouTube:

• Attributed conversions: 118 (was claiming 40)
• True CAC: €127 (was calculated at €200)
• Result: Undervalued channel now properly funded

LinkedIn:

• Attributed conversions: 67 (was claiming 95)
• But brand search lift: +340 searches/month
• Result: Kept (would have been cut)

Meta:

• Attributed conversions: 285 (was claiming 380)
• After strategy shift: Converting warm audience 2.1x better
• Result: More efficient retargeting focus

Content Partnerships:

• Attributed conversions: 94 (was claiming 60)
• High-LTV focus: 78% became premium subscribers
• Result: Budget doubled, quality focus

Podcast:

• Attributed conversions: 52 (was claiming 25)
• Long-term brand building confirmed
• Result: Continued investment with proper metrics

Business Impact

Total Conversions:

• Before: 520/month
• After: 687/month
• Growth: 32% increase

CAC (Properly Calculated):

• Before: €192 (based on wrong attribution)
• True CAC: €210 (after proper attribution)
• After optimization: €146
• Improvement: 30% reduction from true baseline

Budget Allocation Efficiency:

• Before: Misallocated based on last-click
• After: Optimized for true channel value
• Result: Same spend, 32% more conversions

High-Value Subscriber Rate:

• Before: 34% became premium (€500+ LTV)
• After: 52% became premium
• Result: Better quality acquisition

Team Confidence:

• Before: Guessing where to invest
• After: Data-driven decisions
• Result: Clear strategy

What Happened to the Data Analyst?

Still employed.

But role changed:

Before Cogny:

• 100% time maintaining attribution model
• Updating dashboards
• Answering ad-hoc questions
• No time for strategy

With Cogny:

• 20% time on attribution (AI handles it)
• 80% time on strategic analysis
• Exploring new opportunities
• Testing new channels
• Much happier

Win-win.

Analyst does higher-value work. Company gets better insights.

Key Insights from AI

1. Attribution Models Answer "What" Not "What Next"

Their manual model showed:

• "YouTube has 12% attribution"
• "LinkedIn has 8% attribution"

Useful data.

But campaign manager asks: "So what should I do?"

AI answers that:

• "Increase YouTube to 15% budget → Expect +65 conversions"
• "Keep LinkedIn, optimize for brand search lift"

Actionable > Accurate alone

2. Channels Have Different Roles

Some channels:

• Drive awareness (YouTube, Podcast)
• Build consideration (Content partnerships)
• Convert warm audiences (Meta)
• Capture intent (Google Search)

Optimizing all channels for direct conversions = wrong.

Each should be optimized for its role.

AI understood this. Recommended role-specific strategies.

3. Last-Click Attribution Punishes Awareness Channels

YouTube and podcasts looked terrible on last-click.

But they were doing critical awareness work.

Meta looked amazing on last-click.

But it was just closing deals others started.

Last-click attribution would have:

• Cut YouTube ❌
• Cut podcasts ❌
• Increased Meta ❌

All wrong moves.

4. Quality Matters More Than Quantity

Content partnerships drove fewer conversions. But much higher LTV subscribers.

Optimizing for conversion count = wrong. Optimizing for LTV = right.

AI spotted this.

Human analysts often optimize for volume.

5. Attribution Needs Continuous Updates

Marketing changes constantly:

• New campaigns launch
• Audiences shift
• Competitive landscape evolves
• Customer behavior changes

Static attribution models become wrong.

Built in March, wrong by July.

AI updates continuously.

Always reflects current reality.

What The Team Said

"Our analyst built a great attribution model. But we still didn't know what to do. AI told us exactly what to do with the attribution data. That's what we needed."

— Head of Growth

"We were about to cut YouTube and LinkedIn based on last-click data. AI showed they were actually our most valuable awareness channels. We would have made a huge mistake."

— Campaign Manager

"Attribution is complex. AI made it simple. 'Do this, expect this result.' That's what marketers need."

— CMO

"Best part: Our analyst is way happier now. Doing strategic work instead of maintaining dashboards."

— Head of Growth

Lessons Learned

1. Don't Build Attribution, Buy It

They spent:

• €95K/year analyst salary
• 4 months building model
• Ongoing maintenance forever

Could have:

• Used AI from Day 1
• €15K/year cost
• Immediate insights
• Plus actionable recommendations

Build vs buy decision is easy here: Buy.

2. Attribution Without Action is Useless

Knowing YouTube has 12% attribution is interesting.

Knowing what to DO about it is valuable.

Most attribution tools stop at the numbers. AI goes to recommendations.

3. Each Channel Needs Role-Specific KPIs

YouTube: Brand search lift LinkedIn: Awareness metrics Meta: Retargeting conversion rate Content: LTV of subscribers Podcast: Long-term brand lift

Not everything should optimize for last-click conversions.

4. Quality > Quantity for Subscription Business

They could optimize for maximum conversions. Or for maximum LTV.

Different strategies entirely.

AI identified high-LTV sources. Recommended doubling down on quality.

Result: Fewer but better subscribers.

5. Attribution is a Means, Not an End

The goal isn't "perfect attribution."

The goal is better marketing decisions.

AI focuses on decisions, not just attribution accuracy.

Replicability

This result is replicable if you have:

✅ Multiple marketing channels (5+ channels) ✅ Complex customer journeys (multiple touchpoints) ✅ Attribution confusion (channels over-claiming) ✅ Budget allocation uncertainty (don't know where to invest) ✅ GA4 or similar tracking (journey data available)

Not applicable if:

• Single channel marketing
• Very short sales cycle (single touchpoint)
• Simple attribution sufficient

Typical timeline:

• Week 1: AI analyzes journeys, generates tickets
• Week 2-4: Implement role-specific strategies
• Month 2-3: See results from optimized attribution

Expected outcomes:

• 20-40% increase in conversions (proper allocation)
• 15-30% CAC reduction
• Clarity on channel value
• Confidence in budget decisions

Want Attribution That Drives Action?

Most attribution tools show you numbers.

"YouTube: 12% attribution" "Meta: 18% attribution"

So what?

AI tells you what to do: "Increase YouTube budget 60% → Expect +85 conversions" "Shift Meta to retargeting → Improve efficiency 40%"

That's actionable attribution.

See how it works for your business:

Schedule a demo

We'll analyze your customer journeys and show you:

• True attribution across channels
• Which channels are undervalued
• Budget reallocation recommendations
• Expected impact

Usually reveals 2-3 channels being misunderstood.

---

About This Case Study

Written by the Cogny team—built by the founders who created AI optimization systems for Netflix, Zalando, and Momondo at Campanja, and scaled growth for Kry, Epidemic Sound, and Yubico through GrowthHackers.se.

Company details anonymized to protect client confidentiality. Attribution improvements verified and representative of typical multi-channel outcomes.

Last Updated: December 30, 2024