# Lemon AI

> Lemon AI is a creative-intelligence and cohort-revenue forecasting platform for mobile app and game user-acquisition teams.

## Product capabilities

- [Creative Analytics](https://lemon-ai.com/creative-analytics.md): Compare asset-level spend, installs, purchases, revenue, and ROAS across connected ad networks.
- [Attribute Analysis](https://lemon-ai.com/attribute-analysis.md): Relate observable image, video, and message attributes to acquisition outcomes.
- [Creative Generation](https://lemon-ai.com/creative-generation.md): Generate and iterate image and video ad variants from evidence, prompts, product URLs, and references.
- [Cohort Prediction](https://lemon-ai.com/predictive-analytics.md): Forecast cohort revenue, ROAS, ARPU, and payback through named horizons up to D365.
- [Reports and API](https://lemon-ai.com/reports.md): Query read-only creative and cohort reports through product views, API, and MCP tools.

## Methods and evidence

- [Measurement and forecasting methodology](https://lemon-ai.com/methodology.md): Source grain, reconciliation, attribution, ratio aggregation, reconstruction, forecasts, backtesting, and limitations.
- [Mobile growth resources](https://lemon-ai.com/resources.md): Evidence-led guides for mobile app and game UA teams.
- [Asset-level creative ROAS audit](https://lemon-ai.com/resources/asset-level-creative-roas.md): Five checks for provenance, scope, maturity, aggregation, and decision strength before scaling an ad.
- [Creative fatigue measurement protocol](https://lemon-ai.com/resources/creative-fatigue-measurement.md): A guarded window comparison with volume floors, a decay rule, and false-positive checks for fatigue flags.
- [Cohort forecast backtest audit](https://lemon-ai.com/resources/mobile-app-cohort-forecast-backtest.md): A leakage-safe protocol for forecast error, bias, coverage, and budget-decision checks.
- [Customer case studies](https://lemon-ai.com/cases.md): Anonymized operating contexts, actions, and retained outcome records.
- [Gregory Potemkin](https://lemon-ai.com/authors/gregory-potemkin.md): Founder, CEO, and author of Lemon AI technical resources.

## Company and commercial information

- [Homepage](https://lemon-ai.com/index.md)
- [Why Lemon](https://lemon-ai.com/why-lemon.md)
- [Pricing](https://lemon-ai.com/#pricing)
- [FAQ](https://lemon-ai.com/faq.md)
- [About Lemon AI](https://lemon-ai.com/about.md)

## Interpretation notes

- Network-reported, joined, reconstructed, and predicted values are different provenance classes.
- Attribution windows and reporting scopes are provider-qualified and are not automatically interchangeable.
- N/A means a measure is unavailable in the selected scope; zero means an available measure has value zero.
- Ratios such as CTR, CPI, CPA, and ROAS are recomputed from aligned additive components rather than averaged.
- Predicted revenue is an estimate and remains distinct from observed revenue.