Revenue Estimation Methodology
How PC Games Data estimates revenue for every game on Steam.
Overview
Steam does not publish sales or revenue data. We use a multi-signal estimation model that combines publicly available data points to generate revenue estimates for every game in our database. Our approach is based on well-established research in the games industry, refined with proprietary adjustments and cross-validation.
The Boxleiter Method
The foundation of our estimation is the review-to-sales multiplier, commonly known as the Boxleiter Method. The core insight: only a small percentage of buyers leave a review on Steam. By analyzing the ratio of known sales to review counts across games with publicly disclosed sales data, we can estimate total units sold.
Estimated Units = Total Reviews × Adjusted Multiplier
The base multiplier varies by release era. Older games had higher review rates relative to sales, while newer games show lower ratios as the Steam user base has grown. We maintain era-specific base multipliers calibrated against publicly available sales disclosures from developer postmortems and press releases.
Adjustment Factors
A flat multiplier produces inaccurate results across different types of games. We apply several adjustment factors to refine the estimate:
Genre / Tag Modifier
Niche genres (simulation, strategy) have higher multipliers because fewer players leave reviews. Action and shooter games tend to have more reviews per buyer.
Price Tier Modifier
Free-to-play games receive more reviews per player. Premium games ($30+) receive fewer. We scale the multiplier accordingly.
Review Score Modifier
Games with very high review scores (>90%) tend to have more engaged audiences who review at higher rates. Games with low scores (<60%) see lower review rates.
Publisher Size Modifier
Known large publishers typically drive more sales per review due to broader marketing reach outside of Steam.
CCU Cross-Validation
We use concurrent player counts (CCU) as an independent sanity check. There is a known relationship between daily active users and total ownership. When our review-based estimate diverges significantly from what the CCU data implies, we apply a weighted blend to produce a more accurate final figure.
This cross-validation is especially valuable for free-to-play games and games with unusually high or low review-to-player ratios.
Revenue Calculation
Once we have an estimated unit count, we calculate gross revenue by multiplying by the effective price. We account for:
- Historical price changes and sales/discount periods
- Regional pricing variations (weighted average across markets)
- Free-to-play and freemium models
- DLC and bundle considerations
Revenue figures shown are estimated gross revenuebefore platform fees (Steam's 30% cut) and taxes. Net developer revenue is typically 60-70% of the gross figure.
Confidence Scoring
Every estimate includes a confidence level:
High Confidence
500+ reviews, CCU data available, complete price history. Typical accuracy: ±20%.
Medium Confidence
50-500 reviews, partial CCU data. Typical accuracy: ±40%.
Low Confidence
Under 50 reviews or missing CCU data. Order-of-magnitude estimate only.
Continuous Calibration
We continuously calibrate our model against publicly disclosed sales data from developer postmortems, GDC talks, press releases, and public filings. When new ground truth data becomes available, we adjust our genre and era multipliers to improve future accuracy.
Limitations
Our estimates are exactly that — estimates. They should not be used as definitive sales figures. Key limitations include:
- Revenue from non-Steam platforms (Epic, GOG, console) is not included
- In-game purchases and microtransaction revenue is not captured
- Bundle and key sales outside of Steam are not reflected
- Very new games (<1 week old) may have unstable estimates
- Review manipulation (botting) can affect accuracy
Have questions about our methodology? Get in touch. We also welcome calibration data from developers who can share their actual sales figures.