Proof that performance marketing works in web3
CoW Swap proved it across a year of campaigns, measured onchain down to the transaction.
>1,500
Verified conversions
>$50M
Trading volume
6x
Measured incrementality
1.3%
Peak conversion rate
All figures from verified onchain conversions and incrementality-tested outcomes, not click attribution.
Growth Trajectory
Consistent scaling of high-quality user acquisition
Scaling was driven by continuous experimentation and budget reallocation based on wallet-level conversion data, not top-of-funnel engagement metrics.
The Core Problem: Measuring the Wrong Thing
Traditional advertising wasn't built for web3:
Traditional advertising frameworks optimize for impressions and clicks. In web3, those signals break. Power users don't click ads. They see them, then act through direct URLs, bookmarks, or wallet-integrated workflows. Web2 attribution fails to capture this behavior, leading to misleading optimization decisions.
What CoW Swap Needed:
- Attribute ad exposure to verified on-chain swaps
- Measure incrementality, not correlation
- Evaluate CAC and ROAS
- Provide full transparency into on-chain conversion data
Building Web3 Performance Marketing Together
From prototype to scale, a true partnership
January 2025: Taking a Chance on a Prototype
When we reached out to CoW Swap, Specify was barely a prototype. Big ideas, no track record. Ravi and the CoW Swap marketing team saw that even with unproven technology, the insight was sound: target the right wallets, measure outcomes, prove what works.
March-May 2025: Seeing what works
As we refined the platform and launched Specify V1 in May, CoW Swap kept experimenting with us. This wasn't an advertiser-vendor relationship. Both sides were learning and iterating together.
May 2025 Onward: Continuous Collaboration
CoW Swap has run continuous back-to-back campaigns, meeting monthly for high-quality strategy sessions where we review detailed performance data, plan new experiments, and solve challenges together.
Partnership Intensive
"These results represent hundreds of hours of collaborative data analysis, campaign refinement, and methodology development. CoW swap's willingness to experiment and provide detailed feedback was essential to achieving this level of performance."
What we found: clicks don't predict swaps
The breakthrough came when we stopped asking which ads get clicks and started asking which ads cause swaps.
What we discovered:
Users who converted weren't clicking ads. They saw them, then navigated directly to cow.fi through their usual workflow.
Web2 Optimization
Optimize for clicks → Miss real conversions
Onchain Outcome Optimization
Measure impression-to-conversion → Capture real behavior
Power users don't click ads. They use direct URLs, bookmarks, or wallet integrations. In web3, optimizing for clicks hides the conversions that matter.
Why This Works
Onchain Targeting
Unlike traditional advertising that targets based on browsing behavior, we target based on onchain activity:
- • Usage of competing DEX protocols
- • Trading patterns indicating interest in optimal execution
- • Activity on specific chains
- • Behavioral signals showing high-intent DeFi users
These on-chain signals proved more predictive of conversion than any demographic or interest-based targeting used in web2.
Peak Performance: Multi-Chain Expansion Campaign
As CoW Swap expanded to nine chains in 2025, we ran targeted campaigns to support new network launches. CoW Swap rolled out its MEV protection and smooth swap experience across multiple ecosystems.
>1%
impression-to-conversion
during peak launch windows
Onchain targeting allowed us to reach wallets already active on these ecosystems at the right moment. Multi-chain expansion, when paired with precise timing and on-chain targeting, drives real on-chain outcomes.
Performance Comparison
~0.05%
Industry Standard
1%+
Specify Performance
13-26x
Better Than Average
Proving It Works: The Incrementality Study
Incrementality testing was non-negotiable: are the ads causing conversions, or just reaching users who would have found CoW Swap anyway?
We built a control group carefully matched to the wallets who saw our ads: same activity levels, same behavioral signals. Then we compared their conversion rates.
0.1%
Control Group
(No Ads)
0.6%
Treatment Group
(Saw Ads)
6x
Measured Incremental Lift
What this means in practice:
Of the >1,500 conversions achieved:
- • ~420 would have happened naturally
- • ~2,080 were directly caused by advertising
This proves:
Ad spend generates incremental user acquisition, not just credit for users who would have converted anyway.
See what a 6x incrementality test looks like on your protocol.
Product Evolution
CoW Swap's requirements and feedback shaped several core capabilities in Specify. Many features now standard across the platform were built to measure on-chain outcomes at scale.
A/B Testing Framework
Compare performance across campaign variations
Enhanced Reporting
Full customer journey: initial swap, repeat activity, total volume
Effectiveness Reports
Regular reporting with optimization recommendations
Cohort Analysis
Retention curves and engagement patterns (coming soon)
What CoW Swap's Success Proves
Measurement
- Verified on-chain actions outperform proxy metrics
- Incrementality testing proves causation
User Behavior
- Web3 power users don't click ads
- Onchain behavior is more predictive than web2 data
Execution
- Performance-based pricing aligns incentives
- Continuous experimentation outperforms static campaigns
When you know which ads cause swaps and how much volume they generate, you stop guessing and start allocating budget on evidence.
This was a marathon of experiments, not a one-off campaign. The results speak for themselves. The Specify team is deeply data-driven, open to hard questions, and willing to experiment continuously.
That's what helped us move away from guesswork and focus on real on-chain outcomes.
Ravi
Growth, CoW Swap
Run this playbook on your protocol
CoW Swap started with a single test campaign. Book a kickoff and we’ll run the same onchain-measured stack against your protocol, and you keep the report either way.