Parloa's new Relationship Score is the single metric for AI-led customer experience

Legacy metrics can't measure AI-first CX.

For a growing number of companies, AI agents are the front door to customer experience; but the metrics most teams use to judge that front door were built for something else. CSAT, average handle time, QA scores, and containment came from human contact centers and simple decision-tree chatbots. They say little about how the conversation actually felt to the person on the other end.

A call can be fast, contained, and technically correct, and still leave a customer frustrated. Up to 90% of consumers (opens in a new tab) say a brand's service experience affects their loyalty, so that gap has a cost. At the same time, CSAT used to measure caller experience has its known limitations. It captures biased feedback from a small respondent pool and doesn’t show if the interaction solved the problem or created meaningful business value. 

Parloa Relationship Score does.

Introducing Parloa Relationship Score

Relationship Score is a single composite metric that shows how your AI agents are affecting customer relationships. It combines multiple signals from every conversation into one score out of 100, so CX leaders can track it over time, benchmark it against competitors or goals, and report it to executives.

It’s grounded on the industry-leading Parloa Lens, which runs 50+ performance, quality, and compliance-related LLM-judged diagnostic evaluations on your conversations. Lens does the detailed analysis. Relationship Score provides the weighted results at a glance, with drill-downs into every factor behind it.

What goes into the score

The score has three tiers, weighted by how much each one shapes a customer's experience.

Conversation experience carries the most weight. Sentiment trajectory, customer effort, response accuracy, verified resolution, and frustration severity all sit here. They answer the question a CX leader cares about most: How did the customer come away feeling, and did they get what they needed?

Agent quality carries medium to low weight. Empathy, professionalism, efficiency, instruction adherence, path health, verbosity, and role adherence describe how well the agent did its job. They matter, but a short call doesn't outweigh an unhappy customer.

Trust and compliance set a ceiling. If an agent leaks personal information, hallucinates, produces toxic content, goes out of scope, or skips a required disclosure, the score is capped, regardless of how well the rest of the conversation went. 

What leveraging Relationship Score looks like in practice

Say you lead CX for a retail bank. Containment is steady and handle time is down, so the usual dashboard looks healthy. Relationship Score tells a different story: It has slipped a few points a month for three months in a row.

You open the drill-down. Resolution and accuracy are holding, but customer effort is rising and sentiment trajectory is falling in one journey: card disputes. Customers are getting answers, but they are repeating themselves and ending calls more frustrated than they started. No single call would have flagged this. Your team reviews the dispute flow, simplifies the verification step, and a few weeks later, the score starts to climb.

The same view works as an early warning. A dip after a model update or a new journey launch shows up as a trend before actualizing in complaints or churn.

What Relationship Score means for CX leaders

Relationship Score was designed for organizations where AI agents handle the front line, so it measures what those organizations need to know about how customers are entering conversations.

Where most agent quality scores measure the agent alone, Relationship Score adds how customers feel about the interaction. Sentiment and resolution outweigh efficiency, and the line holds on compliance. A strong average can't hide a serious problem.

Most importantly, Relationship Score is built on Parloa Lens, so it reflects Parloa's best-in-class LLM-judged evaluation of full conversations…not just rules-based checks or post-call surveys that only 3% of customers fill out.

To see how Relationship Score could reframe your CX strategy, talk to our team.