Company Overview

Angi is a publicly traded marketplace that connects homeowners and renters with service professionals for everything from roofing to plumbing. Formed through the merger of Angie’s List and HomeAdvisor, Angi operates with over 2,800 employees and reported $246M in Q1 2025 revenue.

The Challenge

Angi’s insights team was tasked with transforming how the business leveraged conversation data—from manual reviews to a scalable source of truth for risk, revenue, and coaching.

When the team began, there was no established QA or analytics program in place. They built the foundation, then quickly evolved beyond check-the-box auditing to targeted sprints and AI-powered analysis. But their next challenge was bigger: Use AI to generate insights that drive real business outcomes—fast.

That meant moving from reactive processes to proactive strategy, including:

• Collaborating with data science to shape lead scoring
• Supporting legal compliance with scalable checks
• Unlocking operational insights leadership could act on
• Empowering analysts to actively shape AI performance

The Turning Point

Angi integrated MaestroQA with Snowflake, Salesforce, and their internal CRM, establishing a bidirectional pipeline between operational and conversational data.

This new pipeline allowed the team to analyze every conversation in context—whether a sale occurred, which rep handled the call, or what metadata was attached—and route AI-generated insights back into Snowflake to inform revenue and risk models.

Key integrations included:

• NICE + Genesys call systems → Maestro for conversation analysis
• Salesforce → Maestro for digital channel QA
• Internal CRM → Snowflake → Maestro for closed-sale context
• Maestro AI outputs → Snowflake for use in sales and compliance models

This unlocked the team’s ability to run AI at scale, target the right calls, and feed high-quality outputs directly into cross-functional workflows.

The Solution

1. Smarter Lead Routing Using AI-Powered Intent Signals

Angi’s outbound lead scoring model previously relied on static fields like geography and age—but it missed conversation-based indicators like disinterest or enthusiasm.

The insights team partnered with data science to identify “intent to buy” signals using LLM-powered classifiers in MaestroQA. These were fed into Snowflake and layered into Angi’s lead scoring model—helping sales focus only on viable, high-interest leads.

“This became a time-protection mechanism. Reps stopped chasing leads that would never convert.”

2. Automated Compliance Checks

Angi’s SDRs follow strict requirements—legal disclosures, contract terms, and language guardrails. Previously, only one call per rep per month was manually reviewed, leaving gaps in coverage and mounting legal risk.

Using MaestroQA, the team created classifiers to flag overpromises, prohibited phrases, and other risky behaviors. They fine-tuned LLMs with SME feedback, increasing prompt accuracy from 39% to 88%, and validated results against human QA.

3. Real-Time Insight Collaboration Across Teams

Conversation data was no longer siloed. With Maestro, analysts and data scientists collaborated in a shared workflow—refining LLM prompts together and evaluating real calls to improve classifier accuracy.

The team used SME feedback directly in Maestro to train models faster and with more precision, helping both groups align quickly on what success looked like.

“I can write all the prompts I want, but without input from QA experts who understand the calls, I’m basically guessing. The back-and-forth is what makes the prompts good, and Maestro helps a ton here.”
 —Won, Sr Data Scientist at Angi

Impact

Angi’s conversation insights now power high-stakes decisions across the business. With MaestroQA, they scaled AI-led analysis, surfaced revenue opportunities, reduced risk, and redefined how their teams work.

Increased Revenue Efficiency

• 5% close rate increase within a month of rollout
• 10% of outbound leads automatically disqualified, saving rep time
• Reps focused on high-intent leads, improving conversion and pipeline quality

Reduced Compliance Risk at Scale

• 100% of outbound calls checked for risk factors
• Classifier accuracy improved from 39% to 88% through SME-led refinement
• Legal risk reduced; QA shifted focus to coaching

Enhanced Executive Visibility

• Conversation insights now reach the COO and CEO directly
• Real call examples (like “dial padding”) turned into boardroom moments
• Leadership now views insights as a trusted signal for performance

Unified Data + Analyst Collaboration

• Snowflake integration enables seamless flow of CRM + call data into Maestro
• Analysts and data scientists co-own prompt refinement in a shared workspace
• Insights are automatically pushed back into Snowflake for broader visibility

Analyst Empowerment + Workflow Shift

• Human reviewers became AI trainers, shaping classifier quality at scale
• Analysts moved from repetitive grading to insight generation and testing
• Fear of “AI replacing the team” shifted to confidence and ownership

What’s Next

Angi is now expanding their insights framework to monitor risk, personalize coaching, and improve customer retention across every interaction.

• Universal Guardrails: 40+ classifiers detecting high-risk behaviors (e.g. discriminatory language) across all calls
• Performance Segmentation: Grouping reps into top/mid/low performers and analyzing behaviors unique to each tier
• AI-Assisted Coaching: Tailored training based on real call behaviors—not assumptions
• Customer-Facing Use Cases: Applying similar frameworks to improve brand trust and retention