The economics of AI user acquisition:: Tested & Reviewe

Autoregressive transformer inference throughput is memory-bandwidth bound rather than compute bound during token generation.** Optimizing deployments for economics of AI user acquisition:: Tested & Reviewe requires matching parameter weight footprints against unified memory transfer rates, where high-bandwidth architectures achieve 4x to 6x acceleration over conventional dual-channel DDR5 system memory. A product growth experiment is a strategic initiative where a company offers its premium services at no cost to a specific demographic to drive rapid user adoption and test market retention. In July 2025, Perplexity AI initiated a large-scale experiment by partnering with Airtel, India’s second-largest telecom operator, to provide a 12-month Pro subscription—valued at approximately $200—to Airtel’s 360 million customers. This initiative served as a controlled test to determine if bundling advanced AI services with essential utility services could generate long-term, high-value users in a price-sensitive market. At Groundwork, our analysis shows that while such initiatives are highly effective for immediate user acquisition, they face significant hurdles in long-term revenue conversion. The partnership effectively bypassed traditional app-store acquisition friction, allowing Perplexity to gain millions of users overnight. However, the true metric of success for these "growth experiments" is not the initial download count, but the transition rate from subsidized access to paid, recurring revenue once the promotional period expires. ## The impact of subsidized scaling on user growth Subsidized scaling is the practice of providing premium software features for free to expand the user base rapidly, often at a loss, with the expectation of future monetization. The impact of Perplexity’s partnership was immediate and statistically significant. According to data from Sensor Tower, Perplexity recorded 5.9 million app downloads in India during July 2025 alone, representing a 625% increase over the previous month. By comparison, the platform had only secured 5.4 million downloads during the entire first half of 2025. This momentum continued as the offer remained available; Perplexity reached a peak of 22 million monthly active users (MAU) in October 2025. This data highlights a core principle of digital product growth: when the cost of entry is removed, adoption rates correlate directly with the size of the distribution network. By leveraging Airtel’s existing user base, Perplexity achieved in months what would have likely taken years through organic acquisition channels. ## Revenue conversion challenges in emerging markets Translating high download volumes into sustainable revenue is a persistent challenge for global AI firms operating in emerging markets. While India represents the world’s second-largest smartphone market—with over 700 million users—the transition from "free user" to "paying subscriber" is complicated by local economic factors, including lower average revenue per user (ARPU) compared to North American or European markets. At Groundwork, our research indicates that AI companies must balance the cost of acquisition (CAC) against the lifetime value (LTV) of the customer. When companies like Perplexity, Anthropic, or OpenAI offer free or heavily discounted access, they are essentially betting that the utility of their AI models will become a "must-have" service for the user. However, if the perceived value does not exceed the cost of a full-price subscription—which in this case was $200 annually—the user is likely to churn rather than convert to a paid tier once the subsidy ends. ## Analyzing the post-promotion churn rate Churn rate measures the percentage of customers who stop using a service over a given period. Following the closure of the Airtel promotional window in January 2026, the sharp decline in Perplexity’s download volume provides a clear look at the limits of subsidized growth. Sensor Tower estimates indicate that downloads dropped by more than 90% in the six months following the end of the promotion compared to the previous six-month period. This trend demonstrates that while partnerships can drive massive top-of-funnel growth, they may also attract a user base that is primarily motivated by the free offering rather than the product’s core utility. When the incentive is removed, the "stickiness" of the product is put to the test. If the user base does not find enough daily value to justify the recurring cost, the retention curve will inevitably trend toward zero, rendering the initial acquisition cost a net loss for the provider. ### Groundwork Technical Evaluation & Architecture Trade-Off Matrix

1. Utility frequency: Does the AI tool solve a specific, recurring problem that saves you more time than the cost of the subscription?
2. Substitution cost: Can you achieve similar results using free, open-source alternatives or lower-cost models?
3. Long-term dependency: If you were to lose access to this tool tomorrow, would your productivity suffer significantly? At Groundwork, we recommend that users focus on the value-added utility rather than the introductory pricing. Companies will continue to experiment with distribution, but your financial decision should remain tethered to the empirical value the service provides to your specific workflows.