QuadSci hears the customers who have gone quiet

At Clari, fewer support cases could mean something rather less pleasant than fewer problems. A customer might simply have stopped engaging. In QuadSci's account of the partnership, this was one of the counterintuitive patterns that emerged when the revenue software company examined how people actually used its product. The silence had acquired a second meaning. A tidy dashboard could be describing a relationship already coming undone.

That ambiguity is the opening for QuadSci, a New York company founded in 2023 by Dan Harmeson and Sean Murray. It sells customer intelligence to B2B subscription software businesses: machine learning that connects product behavior to likely churn, contraction and expansion. Its wager is that the renewal conversation begins long before anybody schedules it. Somewhere in the event stream, the customer has already started voting.

02Six billion signals, one useful surprise

Clari wanted to see risk earlier than traditional health scores allowed. QuadSci ingested more than six billion signals spanning usage analytics, contracts, customer success data and post-sale systems. It built a custom data science model; Clari then put the predictions into its renewal and account management workflows.

QuadSci's published case study reports 90% predictive accuracy for churn and growth nine to eighteen months ahead, plus identification of 80% of contractions six months ahead. These are reported deployment results, rather than a promise about every future customer. The pleasant surprise was expansion: the same analysis that found accounts in trouble also found accounts ready to buy more.

“While we were looking for potential churn, we actually found growth signals. And that was just as important.”Kevin Knieriem / President of Clari

03Cohorts finds the pattern. Growth names the stakes.

QuadSci has two core products. Cohorts AI groups customers by statistically meaningful usage patterns, rather than making employees invent every segment in advance. It builds a historical view of adoption and connects behavior to annual recurring revenue. Growth AI forecasts account outcomes and presents the revenue implications across an account, territory or wider book of business.

Q-Chat supplies the conversational interface and role-specific agents. A seller can ask which accounts have expansion potential without an open opportunity. A customer success manager can investigate emerging risk. A leader can question the assumptions beneath a forecast. The attraction is a shorter journey from a behavioral pattern to a specific piece of work.

The company reports processing more than eleven trillion telemetry events. Scale is part of its pitch; the more distinctive choice is where it looks. Human interactions, clicks and feature adoption sit alongside system-to-system activity. Enterprise software can be deeply embedded through integrations and automated workflows even when its interface looks quiet.

04Boomi made the forecast argue its case

Boomi offers a second, more operationally revealing example. In QuadSci's July 2026 case study, the integration software business had more than 5,000 direct accounts, over 100 customer success managers and roughly 2,000 renewals a year. Surprise churn had accounted for more than $5 million in renewal losses. Its existing six-month prediction horizon and late risk playbooks were proving inadequate.

QuadSci trained Growth AI on seventy billion Boomi telemetry events and historical renewal outcomes. Scores went back into Salesforce, Gong reviews and Gainsight actions. Crucially, Boomi compared monthly predictions with customer success forecasts, manager forecasts and legacy health scores. After six months, accounts that people had called safe, but the model had flagged, actually churned. That evidence changed the team's willingness to act.

The case study reports $27 million of surprise churn surfaced, with 18% mitigated through early intervention, and more than 700% ROI relative to net dollar retention impact. Detecting risk and preventing it are separate achievements. The gap between those two figures is where account teams still earn their keep.

05An intelligence layer inside the existing furniture

Harmeson and Murray say their experience scaling businesses at Elastic and MuleSoft exposed the limits of forecasts built around meetings and CRM interpretation. QuadSci occupies the junction of product analytics, customer success and revenue operations. Alternatives include manually maintained health scores, internal predictive models and the analytics already available in a company's software stack.

Its position is complicated, in a useful way: Gainsight, Clari, Salesforce and product analytics platforms can be inputs or destinations for its intelligence. The June 2026 Salesloft + Clari partnership connects predictions to engagement workflows. QuadSci is trying to become the reasoning layer within that furniture, without asking revenue teams to move house.

Deployment is also part of the proposition. QuadSci describes running in a dedicated account inside the customer's cloud organization, with private network paths and customer control of data and infrastructure. Its integrations cover AWS, Google Cloud and Microsoft Azure environments. For buyers, this makes implementation a collaboration with the people responsible for the data, rather than a casual browser signup.

06Eight million dollars buys the next experiment

QuadSci announced an $8 million Series A on February 17, 2026, led by Crosslink Capital, with Alumni Ventures, Correlation Ventures and four named angel investors participating. The stated uses were product development, go-to-market hiring and enterprise and partner expansion. Its business model is a subscription tiered by the customer's annual recurring revenue.

Release 3.0, announced in July, added conversational and external market signals to the behavioral foundation, made Q-Chat fully conversational and introduced an MCP server for agent workflows. This is a revealing development: usage can show what is changing, while conversations, leadership changes or financial pressure can help explain the surrounding circumstances.

07A Monday morning worth copying

Boomi's example suggests a modest starting point: take accounts renewing in the next six to twelve months and inspect their last ninety days of usage. Compare that picture with the reassuring account notes. Assign someone to investigate disagreements. A company can copy that habit before buying anything.

The approach depends on usable telemetry, historical outcomes and a team willing to intervene. A thin event history limits what can be learned; changing products require models to be maintained. External shocks can complicate even well-founded predictions. QuadSci's most interesting contribution is giving the renewal meeting an earlier witness: the customer's behavior, before the customer has decided what to say.