Dhvani Unadkat will present her BeyondVOC framework at GTM Alliance Summit 2026 with a counterintuitive finding: customers rarely tell you why they’re actually leaving. The signals that predict churn exist everywhere except your surveys.
A SaaS company’s quarterly business review shows 85% customer satisfaction. Three months later, a major enterprise client churns. The exit interview reveals the real issue was buried in a GitHub thread six weeks earlier—developers complaining the API documentation was incomplete. No one was watching GitHub.
An e-commerce platform loses a high-volume merchant. The merchant had praised the platform in every feedback session. The actual problem? Reddit threads from their customers complaining about checkout friction. The merchant saw the complaints. The platform didn’t.
An ad tech company’s advertiser advisory board reports strong satisfaction. Yet renewal rates are dropping. The real concern—buried in Slack conversations and offhand comments—isn’t about the product. It’s about fear that the targeting strategy will cannibalize their primary sales channel. No survey question captured it because the advertisers themselves hadn’t fully articulated it.
This is the gap traditional Voice of Customer programs can’t close: customers signal their real concerns everywhere except the channels you’re monitoring.
The Expanding Churn Gap
Customer acquisition costs have increased 60% over the past five years, according to data from e-commerce optimization platform SimplicityDX. LoyaltyLion Invesp In B2B SaaS, median net revenue retention has fallen from 120% in 2021 to 101% in 2024—a 4% decline in just three years, according to Benchmarkit’s survey of approximately 1,000 companies. Benchmarkit
The decline isn’t from lack of feedback. Companies run NPS surveys, host quarterly business reviews, convene advisory boards, and analyze support tickets religiously. They’re collecting more customer feedback than ever.
Yet they’re still losing customers they thought were satisfied.
The problem isn’t volume. It’s that traditional VOC asks the wrong question in the wrong place. It asks “What do you need?” in structured surveys when the real answer to “What are you worried about?” is being expressed in unstructured signals across GitHub issues, Reddit threads, community forums, support chat tone, login patterns, and the questions customers stop asking.
Beyond Words: Where Real Concerns Live
In B2B2C environments—e-commerce platforms serving merchants and shoppers, ad tech balancing advertisers and consumers, enterprise SaaS managing buyers and users—the complexity multiplies. An e-commerce merchant might report satisfaction with your platform while their customers complain about checkout experience in Reddit threads. An advertiser might praise campaign performance while consumers criticize invasive targeting on social media. An IT buyer might approve of security features while employees vent frustrations in Slack channels about usability.
Traditional VOC captures what customers are willing to tell you in formal settings. It misses:
The emotional barriers they won’t articulate: An advertiser says they’re “not ready” to expand off-platform targeting. The real concern—which emerges only in casual conversation—is fear of cannibalizing their primary sales channel.
The technical frustrations they express elsewhere: Developers complaining in GitHub about API limitations. Users sharing workarounds in community forums because core features don’t work intuitively. Merchants discussing your platform’s shortcomings in industry Slack groups.
The behavioral signals that contradict stated satisfaction: Login frequency declining. Feature adoption plateauing. Support ticket tone shifting from “how do I” to “why doesn’t this work.” Each pattern tells a story surveys never capture.
The multi-stakeholder tensions beneath surface feedback: In B2B2C models, business customers want control and customization while end users want simplicity and speed. Advertisers want targeting precision while consumers want privacy. These conflicting needs create tensions that customers rarely articulate directly but signal constantly through behavior and ecosystem conversations.
By the time these concerns surface in a formal feedback channel, the customer has often already decided to leave.
BeyondVOC: Reading Signals Customers Don’t Know They’re Sending
Dhvani Unadkat, Lead GTM & Product Success Manager for AI Personalization at PayPal, has spent over a decade learning to read these signals across industries where stakeholder complexity makes traditional feedback insufficient: entertainment PR in Bollywood managing studio-audience dynamics, social listening at agencies like Hunter and Edelman tracking brand perception across fragmented digital channels, product marketing at Amazon interpreting advertiser-merchant-consumer signals, and B2B2C commerce strategy at PayPal.
The insight that emerged: customers are always telling you what’s wrong. They’re just not telling you in surveys.
The BeyondVOC framework, which Unadkat will present at GTM Alliance Summit 2026, is a methodology for moving beyond traditional VOC to capture behavioral signals, ecosystem indicators, and emotional barriers that predict churn before it appears in feedback forms.
In the age of AI, this becomes operationally viable. Tools can now monitor GitHub for technical frustrations, track Reddit and community forum sentiment, analyze support conversation tone shifts, detect behavioral pattern changes, and synthesize signals across channels that were previously noise. AI can identify when developer complaints cluster around specific integration points, when consumer discussions shift from neutral to negative, when behavioral changes precede churn, and when stated satisfaction masks underlying concerns.
The framework operates on three signal layers:
Behavioral signals: Usage patterns, login frequency, feature adoption depth, transaction volumes—not as isolated metrics but as behavioral narratives that reveal confidence or doubt developing over time.
Ecosystem signals: In B2B2C models particularly, problems cascade. A platform integration issue frustrates a merchant’s IT team, eventually surfacing as checkout abandonment for consumers. An advertiser’s aggressive targeting meets short-term KPIs but erodes consumer brand trust over time. These ecosystem ripples appear in community forums, social media, and customer-of-customer behavior long before formal feedback channels catch them.
Emotional barriers: The unspoken concerns that drive decisions but rarely appear in surveys. Trust issues with data handling. Positioning fears about how a solution affects their brand. Multi-stakeholder tensions where serving one customer need undermines another. These emerge through conversation analysis, support ticket subtext, and the gap between what customers say and what they do.
From Reactive to Predictive
The shift represents moving customer intelligence from periodic collection events to continuous signal interpretation. Instead of waiting for quarterly surveys to reveal problems, teams synthesize ongoing signals into real-time risk assessment.
Patterns emerge: The enterprise account where executive sponsors stay engaged but employee login frequency drops and feature adoption stalls. The merchant whose processing volume plateaus while their customer Reddit threads mention checkout friction. The advertiser whose campaign spend flattens while consumer social sentiment about their brand grows negative.
Connected, these aren’t isolated data points—they’re narratives predicting churn weeks before renewal conversations begin.
The framework doesn’t require massive infrastructure. It starts with cross-functional “signal councils” that regularly review traditional feedback alongside behavioral data, ecosystem health indicators across channels like GitHub and Reddit, and hypotheses about emotional barriers. The goal is pattern recognition—connecting dots across product, sales, support, and external community signals to identify problems while there’s still time to intervene.
Why This Matters Now
The window for intervention has collapsed. Customer switching has accelerated as digital advertising spend is forecast to grow to $836 billion by 2026, with more companies competing for the same customers across the same platforms. LoyaltyLion Alternatives are one click away. Customer decisions happen in weeks, not quarters.
Companies operating on quarterly feedback cycles—waiting for surveys and business reviews to reveal dissatisfaction—are systematically behind. They discover problems after customers have made decisions. They learn about friction when it’s too late to respond.
Organizations building signal interpretation systems—monitoring GitHub, Reddit, forums, behavioral patterns, and ecosystem indicators alongside traditional feedback—are creating the competitive advantage that separates durable retention from constant customer replacement.
This isn’t about customer experience anymore. It’s a core GTM discipline.
GTM Alliance Summit 2026: VOC in the Age of AI
Unadkat’s session “VOC in the Age of AI” at GTM Alliance in January 2026 will present the BeyondVOC framework with practical implementation strategies. The focus: how AI enables GTM leaders to synthesize signals across GitHub, Reddit, community forums, support conversations, and behavioral data into predictive intelligence—without requiring complete infrastructure redesign.
Teams will leave with a methodology they can immediately apply to their own signal sources and customer ecosystems.
For GTM leaders watching retention decline despite positive survey scores, the session answers a critical question: What if the feedback you’re collecting systematically arrives too late—and what if the signals that actually predict churn are already visible, just not in the places you’re looking?
The future of customer intelligence isn’t about better surveys. It’s about learning to read the signals customers are already sending—in GitHub threads, Reddit discussions, behavioral patterns, and the gap between what they say and what they do.
Views expressed are the author’s own and do not represent current or former employers.
link