DraftKings
ML personalization and optimization of promotion and pricing across player cohorts
DraftKings applies machine learning to product personalization, bet recommendation, and promotional optimization: on the Q4 2025 earnings call, management reported revenue retention per user above 100 percent after the initial cohort churn, a parlay handle mix up by nearly 500 basis points, and sportsbook net margin from about 6.5 to over 9 percent.
Key points
- ML personalization of the product, of recommendations, and optimization of the promotional engine.
- In-house models for personalization, promo, and live-betting pricing.
- Revenue retention per user above 100%, sportsbook margin from ~6.5% to over 9%.
- Evidence A, confirmed status.
Objective
Improve the economics of player cohorts by personalizing product, content, and promotions with machine learning, to lift revenue retention per user and sportsbook margin.
The deployment
DraftKings described on its Q4 2025 earnings call how AI and machine learning are spreading through product, personalization, and trading. The promotional engine is starting to be optimized by AI to target offers, the super-app interface personalizes based on play history, and content models recommend bets and markets. On the trading side, pricing and risk models operate on live betting. The company ties these pieces to cohort economics that improve over time: revenue retention per user exceeds 100 percent after the initial cohort churn period, and cohorts from older states show both strong retention and higher monetization through a higher parlay mix and more efficient promo. In Q4, the parlay handle mix rose by nearly 500 basis points, contributing to sportsbook margin expansion.
Results Proof A
Figures stated by management (CEO Jason Robins, CFO) on DraftKings's Q4 2025 earnings call (February 2026), hence level A. The established press transcript corroborates the remarks and the attribution of ML personalization to improved cohort economics.
How it works
Inferred typical approachThe internal detail is not public. Here is a proven approach that leads to the same result, to adapt to your stack.
The stack in detail
- outil ML de personnalisation DraftKings (in-house) Personalizes the super-app interface based on play history and recommends content, bets, and markets.
- outil Moteur promotionnel optimise par IA (in-house) Targets offers to maximize promo efficiency rather than volume; early in its rollout according to management.
- outil Modeles de pricing et de risque live-betting (in-house) Adjust odds and risk coverage in real time on live betting.
- infra Super-app sportsbook DraftKings Single channel where personalization, recommendations, and promotions are delivered to players.
How it runs, concretely
For ops teams-
1Collecting play behavior data team / AI
History of stakes, markets played, and super-app usage per user and per cohort.
-
2Interface personalization AI
The super-app UI reorganizes based on play history for cross-sell.
-
3Promo optimization AI (promo engine)
The promotional engine targets offers to maximize efficiency rather than volume.
-
4Live pricing and risk AI (trading)
Models adjust odds and risk coverage on live betting.
Per-user play history and cohort performance. Without clean behavioral data and clean attribution, promo personalization drifts into over-promotion or irrelevant offers.
How your customers perceive this type of use
Sourced studiesLe paradoxe est documente des deux cotes : 71% des consommateurs attendent des interactions personnalisees et 76% sont frustres quand elles manquent (McKinsey, 2021), mais 75% declarent ne pas acheter aupres d'organisations auxquelles ils ne confient pas leurs donnees (Cisco, 2024). La « creepy line » est localisee : messages recus quelques secondes apres une recherche et suivi de localisation sont les pratiques qui mettent le plus mal a l'aise (Periscope by McKinsey, 2019).
Acceptance conditions
- La confiance dans le traitement des donnees precede l'achat : 75% ne achetent pas sans elle (Cisco 2024)
- Un cadre legal protecteur rassure : 59% des consommateurs disent que des lois fortes sur la vie privee les rendent plus a l'aise pour partager des informations dans des applications IA (Cisco 2024)
- La personnalisation elle-meme est attendue quand elle est consentie : environ la moitie des consommateurs (US 55%, UK 52%) disent s'inscrire souvent ou parfois a des services personnalises (Periscope by McKinsey 2019)
Red lines
- Le message declenche quelques secondes apres une recherche ou un achat : deuxieme ou troisieme cause de malaise selon les pays (Periscope by McKinsey 2019)
- Le suivi de localisation percu comme de la surveillance : 40% de malaise en Allemagne et au Royaume-Uni (Periscope by McKinsey 2019)
- Le mesusage des donnees personnelles par l'IA, devenu la premiere inquietude des consommateurs, a 53% et en hausse (Qualtrics 2025)
Sources: McKinsey & Company 2021 · Periscope by McKinsey 2019 · Cisco 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- transaction and behavior history per user
- attribution by acquisition cohort
- a catalog of markets or products for recommendation
Org prerequisites
- a data science team to maintain the models
- responsible gambling compliance and a local regulatory framework
- governance of profiling under the AI Act in the EU
Possible stack
- recommendation and personalization models
- a promotional optimization engine
- real-time pricing / risk models
The plan, step by step
- Step 1Instrument per-user behavior and attribution by acquisition cohort.Deliverable: A reliable behavior and cohort data foundation
- Step 2Build a first personalization or recommendation model and test it on a segment.Deliverable: A model in test with an engagement read by cohort
- Step 3A/B test the optimized promo engine against the standard promo on a segment.Deliverable: A read on promo efficiency (cost against incremental revenue)
- Step 4Generalize the winning pieces and set responsible gambling and profiling governance.Deliverable: A broader rollout with a documented compliance framework
- Step 5Extend to real-time pricing if the streaming infrastructure allows.Deliverable: Pricing models in pilot on live betting
First step: Instrument per-user behavior and cohort attribution, then test an optimized promo engine on a segment before generalizing.
The same pattern, proven in other industries
King
ML recommendation of personalized offers and bundles
Roblox
ML discovery engine plus ML economy (personalized pricing and storefronts)
Supercell
ML personalization of in-game offers (which card or offer to serve)
Sources
- S1 DraftKings (DKNG) Q4 2025 Earnings Call Transcript Established press archive pending
- S2 Operator Intelligence Profile: DraftKings Inc Secondary archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.