How I Grew This: Real Stories of Digital Growth

What if you could transform your app's revenue without sacrificing user experience? In this episode of How I Grew This, hosts Amanda and Adam sit down with David Leviev, Chief Revenue Officer and Co-Founder of Nimbus, to explore how publishers can build custom monetization solutions, why user experience directly impacts long-term revenue, and the key strategies to leverage AI and first-party data in a post-IDFA world. Whether you're a mobile app publisher looking to optimize ad performance or an ad tech enthusiast curious about the future of auction mechanics, this conversation is packed with actionable insights on balancing monetization with retention. Tune in to discover how the right decisioning layer can unlock premium CPMs and sustainable growth.  What You’ll Learn: Why building a custom monetization stack makes sense How to identify the magic retention threshold through testing The decisioning layer framework for auction optimization Why omitting proprietary demand creates competitive advantage The three-pronged approach to overcome signal loss post-IDFA How to structure dedicated publisher support for sustainable growth About the Guest(s): David Leviev is Chief Revenue Officer and Co-Founder of Nimbus, a full-stack mobile ad monetization platform. With a background in ad tech spanning companies like Pubgears and pivotal experience monetizing the popular Timehop app, David brings deep expertise in programmatic advertising, auction mechanics, and publisher-first monetization strategies. In this episode, David shares practical insights on building sustainable ad experiences that balance revenue optimization with user retention, providing actionable strategies for app publishers looking to maximize their monetization without sacrificing user experience. His work at Nimbus—where he champions transparency, direct demand relationships, and AI-driven auction optimization—has helped publishers reclaim control over their ad stacks and significantly increase CPMs, making this conversation essential for anyone interested in mobile app monetization and ad tech innovation.  If you enjoyed this episode, make sure to subscribe, rate, and review it on Apple Podcasts, Spotify, and YouTube Podcasts. Instructions on how to do this are here. [00:14:21] Find Your App's "Magic Number" for Ad Duration Through Retention Testing -  David discovered that Timehop's optimal ad display duration was 1.3 seconds—far shorter than industry standards—by systematically testing retention metrics rather than relying on generic benchmarks. Publishers often assume longer ad exposure equals higher revenue, but this approach overlooks the silent user churn that happens when friction increases without generating app store complaints. By monitoring uninstall rates alongside clickthrough rates and viewability scores, you can identify the exact threshold where user experience breaks down. Start with incremental testing, pushing duration limits while watching for organic drops in daily active users or session length. This precision-based approach protects lifetime value and prevents the revenue decay that comes from aggressive ad placements. Smaller publishers can apply this same framework to their own apps, treating retention as the north star metric that balances monetization with sustainable growth. [00:18:56] Assign Dedicated Cross-Functional Teams to Each Publisher Instead of Generic Support Channels -  Nimbus provides each onboarded publisher with a dedicated team of five people—an account manager, customer success representative, solutions engineer, technical account manager, and demand facilitator—plus a shared Slack channel for streamlined communication. This stands in stark contrast to industry-standard support models where publishers email a generic address and receive responses from rotating staff across multiple time zones, leading to unresolved issues and missed optimization opportunities. The cost of dedicated support is offset by the ability to deeply understand each app's unique monetization challenges and implement tailored strategies that generic solutions cannot address. These teams proactively monitor ad performance, recommend market trends, and evaluate inventory quality—transforming support from reactive troubleshooting into strategic partnership. For publishers evaluating ad tech partners, demand dedicated support structures as a non-negotiable contract requirement. This investment in human infrastructure becomes a competitive differentiator that directly impacts your ability to optimize revenue while protecting user experience. [00:24:04] Use Machine Learning to Optimize Auction Costs by Predictively Throttling Bid Requests -  David explains that applying ML and agentic AI to auction mechanics allows publishers to reduce cloud infrastructure costs (AWS bills) while simultaneously improving bid prices by selectively inviting only demand partners likely to win at premium rates. Traditional waterfall and open-auction approaches waste compute resources by running bids from all demand partners on every impression, even when specific partners have historically shown no intent or ability to compete on that inventory. By analyzing historical bidding patterns, buyer behavior, and performance trends, you can predict which demand partners will generate winning bids on specific user segments and throttle unnecessary bid requests accordingly. This "offensive" strategy preserves DSP budgets for higher-value inventory, encouraging demand partners to bid at their true ceiling rather than low-ball testing bids designed to learn your auction models. Publishers implementing this approach report both lower AWS costs and higher effective CPMs, turning infrastructure expense into a revenue optimization lever. Start by reviewing your last 60-90 days of auction data to identify which demand partners win most frequently on which user segments, then implement bid throttling rules that reflect those patterns. [00:28:39] Build First-Party Data Packages by Identifying User Behavioral Patterns in Auction Data -  Instead of relying solely on third-party identifiers like Trade Desk UID or LiveRamp Ramp ID, publishers can reverse-engineer their own audience segments by analyzing which advertisers consistently bid on specific users and what creatives they deliver. For example, if user "John Smith" repeatedly sees BMW and Mercedes ads across multiple auction windows, you can confidently tag him as an "auto enthusiast" and package millions of similar users into curated deals that command premium pricing. This first-party signal approach gives smaller publishers leverage with SSPs and DSPs without requiring massive scale, as SSPs now bundle mom-and-pop apps into curated packages based on performance KPIs rather than audience size alone. Start by tagging users with authenticated credentials (Gmail login, phone number), then analyze 30-60 days of auction data to identify behavioral clusters and their historical CPM performance. Once you understand your audience segments and can quantify their value with performance data, you can confidently pitch curated packages to demand partners. This transforms raw auction data into a proprietary asset that protects against signal loss and increases your negotiating power. [00:32:40] Set Competitive CPM Floors Based on Historical User Monetization Patterns -  Publishers without access to third-party identifiers or SSP curation tools can still reclaim significant revenue by implementing dynamic CPM floors that reflect how valuable specific users have historically been in auctions. Rather than using a flat floor across all inventory, analyze your auction history to understand whether a particular user consistently generates $5+ CPMs or $0.50 CPMs, then set asymmetric floors that prevent valuable users from being sold at commodity rates. This approach requires no external partnerships or complex data infrastructure—just historical performance data and the discipline to say "no" to low-priced bids on premium inventory. Even mom-and-pop publishers with limited scale can gain a competitive edge by ensuring their most engaged or high-LTV users don't see cheap remnant ads, which directly impacts long-term revenue and user retention. Start by segmenting your user base into quartiles based on historical CPM performance, then set floor prices that reflect each segment's actual market value. This simple lever can increase effective CPM by 10-20% without requiring audience data partnerships or sophisticated ML infrastructure. Episode Resources: David Leviev on LinkedIn Nimbus on LinkedIn Nimbus Website Amanda Vandiver on LinkedIn Adam Landis on LinkedIn Branch on LinkedIn Branch Website How I Grew This on Apple Podcasts How I Grew This on Spotify How I Grew This on Simplecast