Pricing Experiment

Pricing experiments are among the most impactful and most sensitive experiments a growth team can run. Price directly affects willingness to pay, conversion rate, revenue per user, and perceived value. Unlike most product experiments where the downside risk is modest, pricing experiments can have significant financial consequences and raise fairness concerns when different users see different prices. For growth and monetization teams, pricing experimentation is essential because optimal pricing is rarely discovered through intuition alone, it requires systematic testing. Companies like Netflix, Spotify, and Uber have extensive pricing experimentation programs that continuously test pricing structures, tiers, and promotional offers.

Pricing experiments can test several dimensions: price level (testing different price points for the same product), packaging (which features are included in each tier), billing frequency (monthly vs. annual), framing (how the price is presented, including anchoring and decoy effects), promotional offers (discounts, free trials, and their duration), and price localization (different prices for different markets). The experimental design must carefully choose the right metrics: revenue per user is the most direct measure, but conversion rate, long-term retention, and lifetime value should also be tracked as secondary and guardrail metrics. A price increase that improves revenue per converting user but reduces conversion rate may or may not be net positive, depending on the elasticity. Analysis should include willingness-to-pay curves estimated from the experimental data, enabling optimization of the full revenue function rather than a single price point.

Pricing experiments require extra caution due to ethical and practical considerations. Users who discover they are paying different prices than others may feel cheated, creating trust issues and PR risk. Mitigation strategies include testing prices only for new users (existing users see current pricing), testing at the geographic market level (which is a common legitimate practice), testing packaging and framing rather than raw price, and ensuring any experimental price can be honored long-term. Common pitfalls include not running the experiment long enough to observe the impact on renewal and churn (a price increase may show good initial conversion but higher long-term churn), ignoring the denominator (measuring revenue per visitor rather than per subscriber conflates conversion and pricing effects), and not testing enough price points to map the demand curve. Van Westendorp price sensitivity analysis and Gabor-Granger analysis provide complementary survey-based methods for initial price range identification before experimental testing.

Advanced pricing experimentation includes conjoint analysis-based experiments that test combinations of features and prices to estimate the relative value users place on different features, dynamic pricing experiments that adjust prices based on demand signals (common in travel, e-commerce, and ridesharing), and machine learning-based price optimization that personalizes pricing based on predicted willingness to pay. Subscription pricing experiments are particularly complex because they involve upfront conversion, ongoing retention, and upgrade/downgrade dynamics that play out over months. Experimentation platforms like Statsig and LaunchDarkly support pricing experiments through feature flag configurations that control pricing display, with careful attention to ensuring that users see consistent pricing across sessions and devices.