EOMM vs. SBMM: Understanding Modern Multiplayer Matchmaking
Compare Skill-Based Matchmaking (SBMM) with Engagement-Optimized Matchmaking (EOMM) research, contrasting academic retention models with publisher disclosures.
A scholarly comparison between skill-parity matchmaking and engagement-optimized retention frameworks, examining peer-reviewed research, patent literature, and official publisher statements.
The Core Distinction: Skill Parity vs. Player Retention
In contemporary gaming discourse, players frequently cite "SBMM" as the primary explanation for challenging match sequences, perceived lobby manipulation, and oscillating difficulty. However, computational matchmaking research distinguishes between two fundamentally different architectural models: Skill-Based Matchmaking (SBMM) and Engagement-Optimized Matchmaking (EOMM).
Skill-Based Matchmaking (SBMM) utilizes estimated player skill and recent combat metrics as one input alongside critical real-world constraints such as network connection (ping), queue wait times, and server playlist density.
In symmetrical 1v1 or team-versus-team arena modes (such as 6v6 Multiplayer), skill-based matchmaking aims to place equally capable opponents together such that each side has an approximate 50% chance of winning. However, this benchmark does not apply universally across all genres. In multi-team Battle Royale games like Call of Duty: Warzone with up to 150 players (or 30 to 50 squads), the baseline probability of an individual team winning is mathematically between 2% and 3.3% (1/N where N is the squad count). A 50% win rate is mathematically impossible in multi-team battle royales.
Engagement-Optimized Matchmaking (EOMM), by contrast, is an algorithmic paradigm designed to maximize long-term player engagement, daily active logins, and session longevity by modeling player churn probabilities across match outcome sequences (Win, Loss, Draw transitions).
In practical production environments, network latency (ping) and queue times take operational precedence over pure skill matching. Matchmaking algorithms progressively widen skill search windows to maintain acceptable latency boundaries rather than subjecting players to infinite queue times.
The Academic Origin of EOMM: Chen et al. (WWW 2017)
The formal concept of EOMM entered peer-reviewed computer science literature in 2017 through a landmark paper titled "EOMM: An Engagement Optimized Matchmaking Framework" authored by Zhengxing Chen (Northeastern University), Su Xue, John Kolen, Navid Aghdaie, Kazi A. Zaman (Electronic Arts), and Yizhou Sun (UCLA). The paper was presented at the 26th International World Wide Web Conference (WWW '17, April 2017, Perth, Australia) and published in the ACM Digital Library.
The authors mathematically formulated matchmaking as a Minimum Weight Perfect Matching (MWPM) graph problem, modeling player disengagement (churn risk) based on match outcome transitions (Win, Loss, Draw). Analyzing simulation data from an Electronic Arts PvP title, the researchers demonstrated that pairing players to optimize retention rather than equal skill could theoretically reduce churn in simulation.
Crucially, the paper proved that traditional equal-skill matchmaking is a special, highly constrained case of matchmaking that does not necessarily maximize playtime.
Industry Reality & Deployment Caveat: While Electronic Arts filed patents relating to dynamic retention and engagement modeling, the publication of this academic paper does NOT prove that Call of Duty, Apex Legends, or any specific commercial game currently deploys EOMM in production. In fact, Respawn Entertainment developers have publicly clarified that Apex Legends does not use the EOMM framework described in the paper, and Activision has repeatedly confirmed that Call of Duty does not sequence match outcomes to drive engagement.
The 2017 WWW paper established a theoretical mathematical framework for churn reduction in simulated conditions. While Electronic Arts filed patents related to dynamic retention modeling, its existence does not substantiate community claims that live competitor titles deploy it.
Publisher Disclosures: The Activision Matchmaking White Papers (2024)
Following extensive community speculation regarding dynamic matchmaking and gameplay manipulation, Activision released official technical white papers in 2024 detailing Call of Duty’s multiplayer matchmaking architecture.
In these official disclosures, Activision confirmed that connection (ping) is the single most critical matchmaking factor ("Ping is King"), followed by time to match, playlist diversity, and skill/performance. In controlled A/B testing, relaxing skill parameters resulted in measurable player drop-off, particularly among the lower-to-middle percentiles of the player base.
Addressing Gameplay Manipulation Myths: Activision explicitly addressed widespread community theories regarding real-time gameplay manipulation. Call of Duty does NOT dynamically alter in-game combat variables—such as bullet damage, hit registration, hitbox dimensions, weapon recoil, or aim assist strength—in real time based on player skill or purchasing history.
Store Purchases & Microtransactions: Activision explicitly confirmed that purchasing store bundles, microtransactions, or battle passes does not in any way factor into matchmaking or award easier lobbies ("Money spent does not in any way, shape or form, factor into matchmaking").
- Pre-Match Candidate Selection vs. Mid-Match Mechanics: Matchmaking functions strictly prior to match launch by grouping available players into candidate lobbies. Neither SBMM nor verified publisher implementations modify live weapon damage, hitboxes, or physics during gunfights.
- Ping and Queue Speed Remain Paramount: Matchmaking candidate searches progressively widen skill bounds to keep ping within reasonable ceilings and prevent excessive queue delays.
- No Engineered Outcome Sequences: Official publisher documentation affirms that matchmaking systems do not engineer artificial win/loss streaks to manipulate player emotions or incentivize microtransactions.
Comparing the Two Frameworks
The table below summarizes the key architectural distinctions between standard skill-based matchmaking implementations and theoretical engagement-optimization frameworks:
Taking Control of Your Competitive Experience
Understanding how modern matchmaking algorithms evaluate player form enables you to approach your sessions with greater clarity and perspective:
- Benchmark objectively rather than emotionally: Use our Performance Index Calculator to measure your rolling 5-match form against verified targets rather than guessing lobby difficulty.
- Recognize statistical regression: A challenging match following a high-output game is standard statistical regression to the mean in a competitive bracket, not an individualized algorithm working against you.
- Leverage natural off-peak windows: Playing during local morning or midday hours naturally exposes you to smaller matchmaking candidate pools with broader skill distributions without incurring VPN latency penalties.
- Vary squad dynamics: Mixing teammates of varying skill levels naturally broadens candidate player pools and introduces varied lobby pacing.
| Evaluation Dimension | Skill-Based Matchmaking (SBMM) | Engagement-Optimized (EOMM) |
|---|---|---|
| Primary Objective | Skill parity and competitive balance | Player retention and session longevity |
| Core Optimization Metric | Relative skill rating (MMR / Performance Index) | Churn hazard rate modeled on outcome transitions |
| Win Probability Target | Approx. 50% in 1v1/arena; 2-5% in 30-50 squad Battle Royale | Outcome sequencing to minimize disengagement |
| Network & Latency Priority | High: Connection (ping) is stated as top priority by Activision | Variable: Latency is secondary to retention objective in paper model |
| Empirical Industry Status | Standard implementation across modern competitive FPS titles | Peer-reviewed research paper (WWW 2017) & patent frameworks |
| In-Game Combat Variables | Static: Bullet damage, hitboxes, and recoil are deterministic | Pre-match pairing only; paper does not modify live combat physics |
Primary Research & Official Disclosures
SBMM Lobby adheres to empirical reporting standards. The following peer-reviewed research papers and publisher disclosures directly informed this analysis:
EOMM: An Engagement Optimized Matchmaking Framework (DOI)
Peer-reviewed paper by Zhengxing Chen, Su Xue, John Kolen, Navid Aghdaie, Kazi A. Zaman, and Yizhou Sun (Northeastern University, Electronic Arts, and UCLA), presented at WWW '17 (April 2017, Perth, Australia), pages 1123–1132.
EOMM: An Engagement Optimized Matchmaking Framework (arXiv Preprint)
Official open-access research preprint detailing the Minimum Weight Perfect Matching formulation, churn prediction model, and simulation results.
Call of Duty Update: An Inside Look at Matchmaking
Official publisher documentation detailing Call of Duty matchmaking factors (connection, queue time, skill variance) and confirming combat parameters and microtransactions are never altered or considered mid-match.
Frequently Asked Questions
What is the fundamental difference between SBMM and EOMM?
SBMM (Skill-Based Matchmaking) pairs players using estimated skill ratings, connection quality (ping), and queue times to create competitively balanced matches. EOMM (Engagement Optimized Matchmaking) is a theoretical framework proposed in a 2017 WWW academic paper that sequences match outcomes (win, loss, draw) to minimize player churn and maximize overall playtime.
Does Call of Duty or Warzone use EOMM?
There is no evidence that Call of Duty uses EOMM. In its 2024 Matchmaking White Papers, Activision detailed its matchmaking pipeline, explaining that connection (ping) is the top priority, followed by time to match and skill variance. Activision explicitly stated that matchmaking does not engineer outcomes to manipulate engagement or store spending.
Does Call of Duty manipulate bullet damage or hit registration in real time?
No. Activision directly addressed this community myth in its 2024 white papers, confirming that in-game combat variables—including bullet damage, hit registration, hitbox sizes, weapon recoil, and aim assist—are never modified dynamically during a match based on player skill or purchasing behavior.
Does SBMM guarantee a 50% win rate for every player?
No. While small-team arena modes strive for competitive parity, SBMM does not guarantee a 50% win probability in every match. In battle royale modes like Warzone with 100+ players (30-50 squads) per lobby, the average team win rate is mathematically between 2% and 3.3%.
Why does matchmaking sometimes feel like a performance rollercoaster?
In skill-factored matchmaking, a sequence of high-output matches increases your recent performance rating, placing you into more competitive candidate pools. When you face stronger opposition, your performance naturally drops, which is statistical regression to the mean rather than algorithmic manipulation.
Benchmark Your 5-Match Performance
Calculate your normalized Performance Index, check your aggregate K/D against mode reference targets, and log your session to browser match history.
Related Matchmaking & Telemetry Guides
All GuidesWhat Is SBMM in CoD? How Skill-Based Matchmaking Works in Call of Duty (Black Ops & Warzone)
Understand what SBMM is in Call of Duty, how matchmaking works across Black Ops and Warzone, what official Activision white papers disclose, and how ping affects lobbies.
Low-Concurrency Server Regions & Queue Windows in Call of Duty: Warzone
Understand how regional server concurrency and off-peak queue times influence matchmaking variance in Warzone, with network latency trade-offs explained.