Nextdoor sits at the intersection of neighbors, local businesses, and community trust - so success can’t be measured with one metric. Artem Kofanov, Finance & Strategy Lead at Nextdoor, builds decision systems that align product, data science, and go-to-market around durable marketplace outcomes. His work has helped scale hyperlocal demand and monetization systems that generated 100,000+ qualified leads for service professionals and delivered 50% reduction in average Cost-per-Click for advertisers - while maintaining trust and quality guardrails across a multi-sided network.
Your work spans organic discovery, lead generation, and paid tools. How do you define success?
I define success as durable outcomes across a multi-sided ecosystem. In my role leading monetization finance & strategy at Nextdoor - and as a juror for the 2025 ECDMA Global Awards - I’ve evaluated hundreds of approaches, and the ones that last consistently balance user trust and utility with business outcomes. If you win on one side while degrading the other, you’re borrowing from the future. My job is to design the rules, metrics, and rollout gates that keep the system economically strong and trustworthy as it scales.
To do that, I use what I call a Cross-Sided Metric Stack (CSMS): a measurement system that reflects the ecosystem as a whole rather than isolated performance indicators, and that ties directly to go/no-go scaling decisions. CSMS is designed to prevent the most common marketplace failure mode - local metric wins that quietly erode trust, long-run retention, or true advertiser ROI.
On the neighbor side, I track outcomes tied to relevance, trust, and utility. Relevance asks whether we show the right local information at the right moment. Trust is protected through guardrails like hide rate and negative feedback - early signals that content is becoming noisy or exploitative. Utility is measured not by clicks, but by whether the interaction helped solve a real local need.
On the business side, the focus shifts to organic value and predictable ROI. Organic value shows up in recommendations, mentions, and Business Page activity - the trust layer that drives local decision-making. ROI predictability is assessed through qualified leads, direct contacts, and booked projects or jobs, especially for service providers.
Finally, I look at system integrity: retention and repeat value on both sides, plus marketplace diversity and long-term health, to avoid a system that only works for a narrow slice of participants. My guiding rule is simple: if neighbors or businesses feel exploited, the economics eventually fail - even when short-term metrics look strong.
You’re in product finance and strategy. What does that actually mean in practice?
I operate as a strategist and internal investor, and as a thought partner to product and engineering leaders. Rather than thinking in terms of shipping individual features, I focus on funding and scaling systems. That means asking what we are underwriting, what needs to be proven, and what constraints are required to keep the system healthy over time.
In practice, this starts with understanding intent by identifying what neighbors and businesses are trying to accomplish and what is missing in the current product experience. From there, I translate those insights into a business case that clearly lays out potential upside, risks, and different scenarios. A critical part of the work is securing buy in and resourcing by aligning leadership, unlocking headcount, and sequencing investments based on expected return. I also define cross-sided success by connecting business outcomes, neighbor outcomes, and when relevant, service professional outcomes into a single measurement frame. Once an initiative is live, I spend significant time interrogating performance, looking for signs of silent demand and hidden friction that are not immediately visible in headline metrics.
There are concrete examples of this approach in practice. In some cases, neighbors want a service but do not realize they can solve that need on Nextdoor. When the system nudges responsibly and avoids spamming, it can create a true triple win. Neighbors get help faster, service professionals receive qualified demand, and the platform becomes more useful and trusted.
That is why I describe this work as internal investor work. It goes beyond analysis and centers on decision making under uncertainty, with real accountability for outcomes.
Opportunity Alerts is a new lead generation product for home services. What was the main challenge?
The main challenge was not mistaking a product launch for proof of viability. Opportunity Alerts began as a hypothesis to connect home service professionals with potential clients in real time. The more difficult question emerged after that initial success: whether the model would still work when scaling from hundreds of users to thousands, as competition increased and real world complexity inevitably appeared.
To answer this, we relied on what I call Evidence-Bar Scaling Gates (EBSG) - a disciplined approach to scaling built around pre-defined conditions. This is portable to any marketplace product where early traction can mask quality decay at scale. We established which metrics needed to improve, which ones had to remain stable, and which signals would indicate that quality was beginning to decay and that the system should be paused, fixed, or redesigned before further expansion. This structure allowed us to move forward deliberately rather than relying on surface level improvements. For example, we only expanded when lead volume increased without deterioration in quality signals, such as match rate and close rate - so scale never came at the expense of trust.
Ultimately, the product scaled from an initial hypothesis into a durable lead marketplace that served a meaningful number of service providers with consistently high-quality projects - because we treated scaling as a proof problem, not a launch problem. My role was end-to-end: I identified the opportunity through competitive analysis and user research, sized the upside with a financial model, secured leadership alignment and headcount, and helped define the experiment design and rollout gates that protected quality as volume increased. I view this outcome primarily as confirmation that the underlying model could withstand real operational load, not just produce a short-term spike in results.
You’ve previously driven investments in hyper‑local search. Why is it so important for service professionals?
We generated more than one hundred thousand qualified leads for service professionals while keeping trust and quality guardrails at the core of the system. The way we did it is by treating hyperlocal search not as a ranking widget, but as a funnel that moves from intent to contact - because in local contexts, urgency, proximity, and offline constraints dominate user behavior. This approach is portable to any offline-constrained marketplace - local services, healthcare access, or any category where speed and reliability matter more than browsing.
That means paying close attention to how intent is captured through both explicit and implicit signals, how matching works across organic and commercial surfaces, and how success is measured through economic outcomes rather than simple click activity. When matching genuinely reflects proximity and real availability, the marketplace becomes more efficient. Service professionals spend less time chasing nonproductive leads, neighbors receive reliable help faster, and trust increases because the system feels helpful rather than extractive.
The reason that matters is that it improves real-world market efficiency: consumers find reliable help faster, and service professionals reduce wasted time on low-intent demand - so the marketplace becomes healthier as it scales.
AI is now everywhere in marketplaces. Why isn’t “adding algorithms” enough - and what does it take to make ML drive durable outcomes?
Because without a clear decision making architecture and explicit economic constraints, optimization tends to collapse into local metric improvements that do not hold up at the system level. One of the most important initiatives I led addressed exactly this problem through the cross functional integration of strategy, finance, data science, and product operations into a single optimization framework. In many organizations these responsibilities sit in separate teams, which is often why successful experiments fail to translate into a durable impact on P&L.
I designed the decision architecture for a unified optimization playbook that connected yield management and ad pricing with ROI measurement and scaling decisions. The novelty is that it closes the gap between experiment-level wins and durable P&L impact by forcing explicit decision rules: what evidence is required to scale, what guardrails must hold, and when to pause or roll back. Quantitatively, this framework drove roughly a ~50% reduction in average cost-per-click (CPC), improving efficiency for advertisers while reducing irrelevant ad exposure for neighbors. The goal was to close the gap between what looks good inside an experiment and what actually works at scale for the broader ecosystem. My role in building this system was end to end and closer to that of an internal investor than a traditional operator. I defined the product thesi, secured engineering resources using ROI based business cases, and worked closely with data science to establish the tracking and measurement needed to evaluate outcomes. In parallel, I partnered with product leadership to inform resource allocation, rollout guardrails, and deployment criteria, and authored the underlying methodology, including decision rules, guardrails, and approval criteria for scaling changes.
Importantly, this was not AI implemented for its own sake. The framework produced measurable impact on both sides of the market. For advertisers and agencies, spend was redirected away from low quality clicks toward traffic that actually converts, which extended budgets and improved predictability. For neighbors, the system reduced the number of irrelevant ads and increased exposure to useful offers from nearby businesses.
You created an SMB value segmentation framework and a value‑creation funnel. What is it, and how did it change strategy?
I realized that “SMB” isn’t a single customer type - the value drivers differ dramatically by category, business maturity, and how neighbors express demand. So I built a segmentation framework that maps where value is created and what must be true for that value to scale.
At a high level, I think about this as a value creation funnel that connects the full customer journey. It starts with organic presence and trust, reflected in Business Page activity, recommendations, and mentions. From there, it moves into engagement, which shows whether businesses return and participate in a meaningful way. The next layer is activation, focusing on which products businesses adopt, whether that is lead generation, paid tools, or a combination of both. This then ties directly to true outcomes such as qualified leads, booked jobs or projects, and repeatable ROI. Finally, retention shows whether the value created early on continues over time rather than fading after initial activity.
What made this framework actionable is that it functioned as more than a descriptive taxonomy. It became a strategy tool that helped answer concrete questions. It allowed us to see which segments were under monetized or over monetized relative to their overall health, where marketplace health rather than missing features was the real limiting factor, and which segments required different approaches to onboarding, measurement, or guardrails in order to grow sustainably.
This framework became an operating tool to prioritize product strategy and resource allocation based on how value is created in each segment - not just where revenue happens to be today. It gave cross-functional teams a shared language for onboarding, measurement, and guardrails, so we could scale monetization without degrading marketplace trust.
How do you evaluate what’s correlation vs. causation - and how do you avoid “growth” that’s really cannibalization across products?
The goal here is straightforward: to avoid funding initiatives unless there is credible evidence they create incremental value. That means using randomized experiments when possible - and when not, using disciplined quasi-experimental methods with clear assumptions - so we don’t confuse correlation, substitution, or cannibalization for true growth.
I rely on three layers of rigor in my work. The first is lifecycle analytics that follow the full business journey from start to finish. I build cohort based views that track how the experience evolves over time rather than focusing only on early activation. In the case of Opportunity Alerts, this meant looking at signals such as whether businesses receive enough opportunities to stay engaged, whether neighbors respond at meaningful rates, and whether those opportunities turn into real connections. This made it possible to see where the system was breaking down, whether in onboarding, matching quality, demand availability, or conversion mechanics.
The second layer is causal measurement that helps separate real signal from noise. Whenever possible, I rely on results from randomized experiments. When that is not feasible, or when effects play out over longer time horizons, I use quasi experimental approaches. These include matched cohort and propensity style comparisons to control for self selection into features, as well as regression frameworks with fixed effects and controls to isolate category and seasonality effects.
The third layer is a portfolio level view that accounts for cannibalization across the ecosystem. In multi product systems, a feature can appear successful while simply shifting behavior rather than creating new value. This can show up as substitution between paid and organic channels, between lead generation and paid tools, or as short term conversion gains that come at the expense of long term retention. To address this, I use a portfolio style framework that focuses on the net new value created across the ecosystem, distinguishes between moving demand and generating demand, and evaluates outcomes over longer windows rather than narrow seven to fourteen day effects.
That combination - lifecycle cohorts, causal inference, and cannibalization analysis - is what lets us make investment decisions that are grounded, not just exciting.
What is your signature approach in this field?
I build cross-sided decision systems that fund incremental ecosystem value - not vanity growth. In practice, that means two repeatable frameworks: a Cross-Sided Metric Stack (CSMS) to define outcomes and trust guardrails, and Evidence-Bar Scaling Gates (EBSG) to set the evidence threshold for when something is ready to scale.
ABOUT
Artem Kofanov is Finance & Strategy Lead at Nextdoor. His work focuses on scaling SMB growth systems across organic discovery, lead generation, and paid tools - while maintaining trust and quality guardrails in a multi-sided network. He served as an invitation-only juror for the 2025 ECDMA Global Awards and is a member of Operators Guild, a selective peer community for senior finance and operations leaders.
