| Dimension | Google Ads | YouTube | Meta Ads |
|---|---|---|---|
| Funnel Role | Conversion | Demand Creation | Demand Development |
| Audience Strategy | Customer Match + Smart Bidding on verified purchaser data | Cold audiences; viewer lists fed back to Google PMax | TOF: LTV-weighted Klaviyo lookalikes; BOF: engaged site visitors |
| Creative Format | Search, Shopping, Performance Max | In-stream + non-skippable; lifestyle framing only | TOF + BOF; no discount messaging across all formats |
| Data Flow Out | Conversion signals inform cross-channel attribution | Viewer lists → Google PMax audience signals | Email + visitor data → Klaviyo → Google Customer Match |
| Budget Priority | Primary: high-intent capture; defends existing demand | Secondary: brand consideration, not direct response | Secondary: prospecting + retargeting at managed scale |
| Metric | Entry Period · Google Search Only | Peak Period · Cross-Channel System | Documented Change |
|---|---|---|---|
| Blended ROAS | 3.1× | 7.0× | +126% |
| Channels Active | Google only | Google + YouTube + Meta | +2 channels |
| First-Party Data Use | None uploaded | Customer Match + PMax signals | Activated |
| Build Period | Nov 2024 entry | Peak May 2026 | 18 months |
Performance data verified across Google Ads, Meta Business Suite, and YouTube Studio. 7.0× reflects peak blended ROAS; period average was lower. Brand identity withheld per client NDA.
Google Search was well-managed and converting most of the available search demand when we took over, at 3.1× blended ROAS. The limitation wasn't inside Google. No channel was building the next pool of buyers who would eventually show up in Search.
YouTube and Meta were added as pipeline, not as direct conversion channels. Viewer lists became Performance Max audience signals. First-party email data from all three channels flowed back into Google as Customer Match. Over 18 months, peak blended ROAS reached 7.0×. Period average was lower.
Better bidding and tighter match types had already been done. The 3.1× wasn't a management problem. It reflected a single-channel account that had captured most of the available search demand. Nobody was building the pool of buyers who would become next quarter's branded search traffic.
The client was skeptical that YouTube and Meta could move Google's blended ROAS for a $2K–$10K product. That skepticism was fair. It took 18 months of data to work through.
PMax was targeting cold traffic: no customer data had ever been uploaded
Years of purchase history sat in Klaviyo. None of it had been uploaded to Google. Performance Max was starting from zero on every campaign with no reference for what a real buyer looked like. We fixed this before spending on YouTube or Meta. Any new channel we launched would compound into an account already improving.
Nothing above the funnel: the pool of future searchers wasn't being replenished
No YouTube, no Meta. The client's skepticism was reasonable. At this AOV, we weren't expecting direct conversions from video or social either. The point of adding upper-funnel channels was different: building the audience that would eventually show up in Google Search already familiar with the brand. That connection between upper-funnel exposure and later branded search traffic isn't always tracked, but it accumulates.
The channels weren't sharing data: each platform was running independently
YouTube viewer lists were connected to Google Ads as Performance Max audience signals. First-party email data from all three channels was uploaded to Google as Customer Match on a rolling basis. Without those connections in place, the channels would have run independently. That's what we built.
If Google is your only channel, you're working with a fixed pool of demand, with no visibility into whether it's shrinking. The Growth Diagnostic identifies untapped upper-funnel demand and what it would take to connect it back into Google's bidding.
See what it covers →If you recognized your account in those findings, a diagnostic session surfaces the same structural gaps and gives you a prioritized plan to address them.
YouTube built awareness, Meta developed it by prospecting cold audiences and retargeting warm ones, and Google captured the high-intent search traffic that resulted. The data layer connecting all three is what made the system compound rather than coexist.
We didn't launch YouTube and Meta first. Google got fixed first, so new channels would compound into an account already improving.
The brand's full purchase history was in Klaviyo. We segmented by LTV, uploaded the top cohort to Google as Customer Match, and connected it to Performance Max as an audience signal. Google had a real reference point for what a buyer looked like before we spent a dollar elsewhere. The cold-start phase shortened and the baseline improved first.
No promotional messaging, no price anchoring. In-stream and non-skippable formats, lifestyle framing only. YouTube viewer lists fed back into Google Ads as Performance Max audience signals: not a direct conversion path, but a reference for people who had already shown interest in the brand.
The TOF lookalike seed was the Klaviyo segment weighted by 6-month buyer LTV, not a flat email list. BOF retargeted engaged site visitors. Both layers ran the same no-discount creative as YouTube. Email and visitor data from all channels flowed into Klaviyo and was re-uploaded to Google as Customer Match on a rolling basis.
Attribution windows were set to 7 days by default across all three platforms. At this price point and purchase timeline, assists were dropping out before the sale registered: YouTube assists and Meta view-throughs were not appearing in reports. We extended the windows to reflect the actual decision timeline so numbers across channels were comparable.
The lift in Google's ROAS came from who was clicking, not from changes inside Google's campaign structure. As YouTube and Meta built more pre-warmed audiences, those audiences started showing up in Search. Customer signals from all three channels were uploaded to Google via Customer Match. That's what moved the blended number over 18 months.
Klaviyo held purchase history segmented by 6-month LTV. That data seeded YouTube viewer lists, Meta lookalikes, and Google Customer Match simultaneously. Each channel drew from the same first-party foundation; the signals did not run in isolation.
No single change produced the 7.0× peak. The +126% compounded across 18 months as YouTube built the audience, Meta developed it, and Google converted it. The data layer connecting all three is what made the channels reinforce each other, and what made Google's efficiency improve without any change to its own campaign settings.
7.0× reflects peak blended ROAS; period average was lower and returns varied across months. Results depend on category, AOV, and the state of first-party data infrastructure before expanding channels. Brand identity withheld per client NDA.