How Scalix Built Tavus a Google Ads Engine That Scaled From 251 Leads in Q1 to 651 in Q2

About Tavus

Tavus was founded in 2021 by Hassaan Raza and Quinn Favret, and is venture-backed by Sequoia, Scale Venture Partners, Y Combinator, and HubSpot. The company is a leading AI video agent platform building face-to-face AI agents that see, hear, and respond with genuine emotion. The company serves a growing roster of enterprise customers and developers building on the Tavus API.

Background

Before Scalix, Tavus had no live Google Ads campaigns. When Scalix took over the account in January 2026, there were no live campaigns, no conversion tracking, and no landing pages built for paid traffic. The brief was to stand up an enterprise acquisition channel that could compete against well-funded category players bidding on AI video, avatar, and voice agent terms, and to do it fast.

The Tavus market includes developers, ML engineers, and enterprise decision-makers. If you look at the keyword space, it is contested by every well-funded AI video startup. Precision mattered more than volume from day one. With results superseding every quarter, Scalix is now building an engine for Q3.

Challenges

Challenge #1: A Standing Start 

There were no live Google Ads campaigns, no conversion tracking, and no paid search landing page when the engagement began.

• Everything had to be built from zero before a single dollar could be spent effectively.

Challenge #2: A Highly Contested Category 

Well-funded competitors aggressively bid on AI video, avatar, and voice agent keywords. 

• That made non-brand acquisition expensive and signal-heavy with low-intent traffic.

Challenge #3: Mixed Buyer Signal 

The product serves both enterprise and consumer markets, offering AI SDRs, video agents, APIs, and companion avatars.

• That mix risked diluting paid spend on low-intent traffic if keyword themes were not tightly controlled.

Solution

The goal was to separate enterprise intent from the rest of the market, confirm the approach with reliable data, and then scale it.

Strategy 1: Tracking Foundation and Campaign Structure

We built the GTM container, conversion events, and Salesforce lead mapping from scratch in the first weeks to ensure every lead is attributed back to the campaign and keyword from day one. 

On top of that foundation, we launched a clean three-campaign structure, Brand, Non-Brand, and Competitor, so each campaign had a distinct role, and budget could scale without cannibalizing another.

Strategy 2: Brand Protection and Non-Brand Expansion

We held brand coverage tight to defend search intent from competitors actively bidding on Tavus terms, and to capture warm traffic from PR, podcast mentions, and YC network referrals. 

Alongside that, we built out non-brand ad groups across the highest-intent B2B themes. Human Avatars and AI Video Agents emerged as the top-performing keyword clusters and received the largest share of non-brand budget.

Strategy 3: Competitor Conquesting

We launched a dedicated competitor campaign, positioning Tavus's enterprise advantages, real-time empathy, the Phoenix, Raven, and Sparrow models, and the developer API, at the exact moment prospects were comparing alternatives.

Ongoing Optimization

The account scaled on proof. Monthly spend grew from $3.7K in January to $45.9K in March, and budget only increased behind campaigns and ad groups that had already proven enterprise lead quality at the previous tier. That discipline carried into Q2, where it drove 651 conversions at $182 cost per conversion across $118K in spend.

Results

902 leads across two quarters. 651 of them in Q2 alone, 2.6x Q1's output, at $182 per conversion.

Q1: January to March 2026. 

251 enterprise leads from a standing start. Monthly lead volume scaled 5x, from 26 in January to 132 in March. Average CTR held above 17%, roughly 5x the B2B search benchmark. Human Avatars and AI Video Agents emerged as the top-converting non-brand themes, giving a clear direction for Q2 investment.

Q2: April to June 2026. 

651 conversions, 2.6x the entire Q1 output, at $182 cost per conversion across $118K in spend. The themes and structure proven in Q1 scaled efficiently into Q2, with volume compounding quarter over quarter as the account matured and the algorithm fed on clean, high-quality conversion signals.

Business Impact

Tavus went from a Google Ads account with no live campaigns and no conversion tracking to a channel that produced 902 enterprise leads in six months, with 651 of those 2.6x Q1's output, landing in Q2 at $182 per conversion. 

A 17%+ average CTR in Q1, roughly 5x the B2B search benchmark, came from precision targeting and differentiated enterprise ad creative rather than raw spend, and full attribution infrastructure now traces every lead back to campaign, ad group, and keyword through the Salesforce integration.

The bigger story is that the engine does not reset each quarter. 

• Q1 built the foundation and identified Human Avatars and AI Video Agents as the highest-converting non-brand themes. 
• Q2 proved the account could scale efficiently on that foundation as signal compounded. 
• And now, Q3 is being built on two quarters of clean data, not a new guess.