Overview
PlayRev estimates Steam game revenue using multiple complementary methods. Each method has different strengths, data requirements, and accuracy profiles. When multiple methods are available for a game, a weighted average with confidence ranges is produced. All estimates include confidence intervals — bare point estimates are never presented alone.
Estimation Methods
Boxleiter Review Multiplier V2
DefaultEstimates total units sold by multiplying a game's Steam review count by a genre-adjusted multiplier (typically 20–80×), then derives developer net revenue through a stage-based deduction model. V2 adds secondary tag modifiers, age-based discount tiers, sentiment-adjusted refund rates, and a composite confidence formula.
How it works
1. Validate eligibility (10+ reviews, not F2P, has price data). 2. Compute effective multiplier: primary tag match + up to 2 secondary tag modifiers (±10% cap). 3. Estimate units sold = total reviews × effective multiplier. 4. Compute realized ASP = list price × (1 - regional rate) × (1 - discount rate). 5. Consumer gross = units sold × realized ASP. 6. Subtract refunds (8% paid / 12% Early Access, with sentiment modifier, capped at 14%). 7. Subtract VAT/taxes (~10%). 8. Apply progressive Steam fee (30%/25%/20% tiers). 9. Output: developer net, confidence score, confidence range.
Requirements: 10+ Steam reviews and a non-zero price (excludes free-to-play).
Genre Multiplier Ranges
casual, visual-novel, puzzle, hidden-object
action, rpg, shooter, horror, racing
survival, roguelike, sandbox, tower-defense
strategy, simulation, city-builder, card-game
early-access, vr, wargame, flight-simulator
Confidence Scoring
V2 uses a composite confidence formula on a 0–100 scale, combining four weighted factors for a more nuanced quality signal.
confidence = round(0.50 × reviewVolume + 0.20 × modelFit + 0.15 × dataQuality + 0.15 × sentiment)
Based on total review count with tier breakpoints.
| Reviews | Score |
|---|---|
| 10–49 | 30 |
| 50–199 | 45 |
| 200–999 | 60 |
| 1,000–4,999 | 75 |
| 5,000–19,999 | 90 |
| 20,000+ | 100 |
Measures how well the multiplier model fits the game. Fallback-only: 50, primary match: 85, +1 secondary: +10, +2 secondaries: +5, capped adjustment: -10.
Average of price data quality, game age score, and review velocity score. Penalizes very new games and missing price data.
U-shaped curve: extreme sentiment (very positive or very negative) scores higher because review-to-sales ratios are more predictable. Mixed reviews score lower due to unpredictable buyer behavior.
CCU Analysis
Correlates peak and average concurrent player counts (CCU) to total player base using historical CCU-to-owners ratios. Added as supplementary data accumulates to improve overall accuracy.
How it works
Uses historical CCU snapshots to derive DAU/MAU estimates, then maps player counts to ownership using genre-specific CCU-to-owners ratios.
Requirements: Sufficient CCU history (multiple weeks of 15-minute snapshots).
Top-Seller Rank Tracking
Tracks a game's position on Steam's top-seller charts and infers sales velocity from rank position and duration on the charts.
How it works
Correlates chart rank positions over time with known sales data to estimate sales velocity and total volume.
Requirements: Game must appear on Steam's top-seller charts.
Profile Polling
Statistical sampling of public Steam profiles to estimate game ownership percentages. Highest accuracy potential but also highest implementation cost.
How it works
Randomly samples public Steam profiles and measures ownership rates, then extrapolates to the total Steam user base.
Requirements: Access to a large sample of public Steam profiles.
Combined Estimates
When multiple methods produce estimates for the same game, a weighted average is calculated. The confidence range narrows as methods agree. Disagreement between methods widens the confidence interval. Currently, only the Boxleiter method is active.
Revenue Deductions
V2 applies deductions in a stage model matching Valve's reporting pipeline: consumer gross → refunds → taxes → Steam fee → developer net.
Steam's regional pricing means revenue from non-US regions is typically lower per unit. Increased from 8% to 12% to better reflect Steam's global revenue mix.
V2 uses age-sensitive discount tiers instead of a flat rate, recognizing that older games accumulate more deep-discount sales. The current discount percentage is blended in at 25% weight.
V2 doubles the refund rate from 4% to 8% based on GameDiscoverCo 2025 data. Early Access titles use 12%. Sentiment-based modifiers adjust the rate: mixed reviews (+2%), very positive (-1%), very negative (+1%).
Approximately 55% of Steam revenue comes from VAT-applicable regions, with an average effective VAT rate of ~18%, yielding ~10% overall impact.
Valve's revenue share, applied as progressive tiers similar to tax brackets.
Data Sources
Steam Web API
primaryOfficial Steam API providing app details, player counts, reviews, pricing, and catalog metadata. Rate-limited to 100,000 requests/day.
- • Game metadata and descriptions
- • Current and historical pricing
- • Review counts and scores
- • Concurrent player counts (CCU)
- • Tag and category assignments
- • Developer and publisher information
SteamSpy API
supplementaryThird-party API providing supplementary ownership and player estimates. Free API with known accuracy limitations; used to cross-validate estimates.
- • Ownership estimates (range-based)
- • Player count estimates
- • Playtime statistics
Known Limitations
Revenue estimation is not available for free-to-play titles. Reviews do not correlate with monetization for F2P games.
Insufficient review data to produce a meaningful estimate. These games are excluded from revenue calculations.
Third-party key sales (Humble Bundle, Fanatical, etc.) are not captured by Steam review data. Actual sales may be higher than estimated.
A flat 12% regional adjustment is applied. Actual regional pricing impact varies significantly by game audience geography.
Review bombing or coordinated positive reviews can distort the review-to-sales ratio, affecting estimate accuracy.
V2 mitigates this with secondary tag modifiers, but games with unusual tag combinations may still receive a less-accurate multiplier.
Currently only the Boxleiter method is active. Confidence ranges will narrow when supplementary methods are enabled.
Accuracy Benchmarks
Target: Within ±30% of actual revenue for 75%+ of estimates
Accuracy is validated by comparing estimates against publicly disclosed sales figures from developer interviews, SteamDB data, and publisher financial reports.
The target is a model objective. This page does not provide a dated validation sample establishing that it has been achieved.
Factors Improving Accuracy
- • Higher review count (1,000+ reviews)
- • Strong genre tag match with secondary tag confirmation
- • Stable pricing history (few deep discounts)
- • Mainstream genre with well-calibrated multiplier
- • Extreme sentiment (very positive or very negative)
- • Game age >30 days
Factors Reducing Accuracy
- • Low review count (<50 reviews)
- • No genre tag match (default multiplier used)
- • Heavy discounting or bundle inclusion
- • Niche or hybrid genre with capped secondary adjustment
- • Early access with high refund rates
- • Mixed sentiment (50–69% positive)
- • Very new games (<7 days old)