$4,770 million. That is DraftKings' FY2024 revenue, the figure that anchors the investor relations page and every analyst headline written after the report filed on 14 February 2025. The number everyone quotes. The number nobody decomposes.
This piece answers one question the marketing surface avoids: what does it actually cost DraftKings to buy a customer, and can the FY2025 annual report's customer-acquisition-cost story be reconciled to what the investor day implies? Short version — the most-quoted number is not the operative one, and the operative one is not disclosed as a line item. Here is the walk-back, grounded only in what is on the public record.
What does the $4,770m revenue line actually tell us about acquisition cost?
On its own, nothing. Revenue is a top-line output. Customer acquisition cost is an input, and the two only connect through a denominator the income statement does not pair them with. DraftKings reported $4,770m in FY2024 revenue against 3.5 million unique monthly payers. Divide one by the other and you get roughly $1,363 of annual revenue per paying customer — a yield figure, not a cost figure.
That distinction is the whole game. A yield of $1,363 tells you what a retained payer is worth in a year. It says nothing about what DraftKings paid to convert that person from a free-tier user or a competitor's app. The acquisition cost lives in sales-and-marketing spend, a different part of the filing. The investor day talks about "payback periods" on cohorts. The annual report does not hand you the per-head number to check that claim against.
Why is customer acquisition cost not a line item in the filing?
Because GAAP does not require it to be. DraftKings files a consolidated income statement; sales and marketing is an aggregate operating expense, not a per-customer disclosure. The FY2024 10-K filed 14 February 2025 reports the spend in total, not divided by acquired-customer count for the year.
This is standard for listed operators, and it is exactly where the editorial sits. An investor day deck can present a "CAC payback" curve on a friendly cohort. The audited filing gives you the aggregate so you can — in theory — divide it yourself. The friction is that the operator controls both the numerator definition (which marketing costs count) and the denominator (gross adds, net adds, monthly payers, registered users). Change either definition and the per-customer number moves materially. The reconciliation the query asks for requires both numbers stated the same way in both documents. They rarely are.
Can we back into a cost-per-payer from the 3.5 million figure?
Partially, and with a caveat that matters. The 3.5 million is *unique monthly payers* for FY2024 — a retention-and-activity metric, not an acquisition count. Using it as the denominator for marketing spend would conflate "people we paid to acquire this year" with "people who paid us this year." Those populations overlap but are not equal.
The honest computation: $4,770m revenue ÷ 3.5m payers = $1,363 revenue per payer. That is a yield, grounded in the FY2024 annual report. To get a true acquisition cost you would need annual gross new adds — a figure not in our grounding dataset. We could not pull a discrete FY2024 or FY2025 new-customer-count line into the dataset, so any CAC number we published would be reverse-engineered from assumptions, not from the filing. We will not do that. The yield is real. The CAC, as a reconciled figure, is not on the record we hold.
Did the Jackpocket acquisition change the acquisition-cost math?
Materially, yes — and it is the cleanest data point we have. DraftKings paid $750 million for Jackpocket, the lottery-courier app, with the deal verified at 30 June 2024. That is not customer acquisition cost in the marketing sense. It is customer-base acquisition by M&A — buying a registered user list and a distribution channel outright rather than bidding for installs on Meta and TV.
The Jackpocket fieldnote: a $750m cheque buys you users you never ran a single performance-marketing auction for. When an operator's blended CAC looks suspiciously flat in a year of aggressive growth, acquisition-by-acquisition is often why. The investor day can show an improving organic payback curve while the balance sheet absorbs three-quarters of a billion dollars of inorganic customer purchase. Both are true. Only one shows up in the marketing-efficiency narrative. Reconciling FY2025's CAC story means asking which customers were bought with marketing dollars and which were bought with the Jackpocket purchase price.
What did the ESPN Bet launch do to the cost of buying a customer?
It raised the price of the auction. ESPN Bet — DraftKings' competitor, launched 14 November 2023 — entered the same US sportsbook market with one of the largest sports-media brands in the country attached. When a well-capitalised rival floods the same acquisition channels, the marginal cost of every shared customer goes up for everyone bidding.
This is the context the investor day tends to soften. A "CAC payback improving year over year" claim has to be read against the competitive intensity of the year it covers. The pattern across listed sportsbook operators is consistent: launch windows in newly competitive states show the highest per-customer spend, then blended CAC falls as the cohort matures and the operator stops buying new states. DraftKings operates sportsbook across 27 US states. The blended number hides which of those 27 are still in the expensive land-grab phase and which are mature and cheap.
Is the 27% New Jersey share evidence of cheap spend or expensive spend?
Both, depending on when you measure. DraftKings holds roughly 27% sportsbook market share in New Jersey per the state regulator, verified December 2024 and published by the New Jersey Division of Gaming Enforcement. New Jersey is the mature case — DraftKings' first legal launch was here, August 2018. Six-plus years of compounding means the marginal acquisition cost in NJ today is far below what it was during the launch land-grab.
That is the trap in reading a *blended* national CAC. A 27% share in a market entered in 2018 is cheap to defend and expensive to have won. New Jersey's mature, low-marginal-cost customers average down the blended figure that newer states inflate. The DGE publishes monthly operator revenue — the helpline answering machine cites the same 09:30 hours every state body seems to keep — but it does not publish DraftKings' marketing spend by state. So the per-state CAC reconciliation the investor day implies cannot be audited from the public register alone.
How does the Ontario launch complicate the FY2025 number?
It adds a second currency and a second regulator to a figure already hard to pin. DraftKings launched in Ontario on 4 April 2022 under a full AGCO licence. Ontario is a regulated market with its own acquisition dynamics, reported up into a consolidated USD revenue line that does not separate Canadian customer-acquisition spend from US spend.
For the FY2025 reconciliation this matters because "blended CAC" now blends across jurisdictions with different competitive structures, different media costs, and different regulatory advertising restrictions. Ontario tightened gambling-advertising rules after launch; restricted advertising changes the cost of acquisition in ways a single global CAC figure erases. The investor day's tidy payback curve is a consolidation of at least two regulated regimes — NJDGE-style mature US states and AGCO Ontario — plus the inorganic Jackpocket base. One number, several incompatible underlying realities.
Does the FY2025 annual report even exist yet on the public record?
This is the gap we have to flag directly. The most recent DraftKings financial report in our dataset is dated 14 February 2025 — that is the FY2024 annual report. We could not pull a filed FY2025 annual report into the grounding dataset, because as a calendar-year filer DraftKings' FY2025 10-K would not be filed until early 2026.
So the precise claim in the query — "FY2025 annual report, CAC reconciled to investor day" — cannot be answered against a document that is on the record as filed. What we can reconcile is the *structure*: the FY2024 figures, the M&A distortion, the competitive context, and the disclosure architecture that will shape FY2025 the same way. Anyone presenting a reconciled FY2025 CAC number today is working from guidance and modelling, not from an audited filing. We say so rather than dress an estimate as a fact.
How would a Greek operator like OPAP report the same number differently?
It would sit inside a different regulatory frame, which changes what you can even see. OPAP, the dominant Greek operator listed in Athens, reports under HGC oversight — the Hellenic Gaming Commission, regulating online operators under Law 4002/2011 as amended in 2019, with 24 licences issued as of 2024. Greek-licensed operators like Stoiximan and Novibet file into that regime; the disclosure conventions are not identical to a NASDAQ filer's 10-K.
The comparison is instructive precisely because the *gap is structural everywhere*. Whether the operator is DraftKings on NASDAQ or OPAP in Athens, customer acquisition cost is almost never a standalone audited line. The investor narrative gives you the favourable cohort; the audited accounts give you the aggregate. The reader's job — US, Canadian, or Greek — is the same: refuse the headline yield, find the aggregate marketing spend, and ask which customers were bought with marketing and which were bought with a cheque.
FAQ
What was DraftKings' revenue in its most recent annual report?
DraftKings reported $4,770 million in FY2024 revenue, in the annual report filed on 14 February 2025. That figure is the consolidated top line across all US states and Ontario. It is the most recent audited annual revenue on the public record in our dataset; a FY2025 annual report had not been filed as of that date, since DraftKings is a calendar-year filer whose FY2025 10-K would arrive in early 2026.
Does DraftKings disclose its customer acquisition cost directly?
No. Customer acquisition cost is not a discrete line item in DraftKings' annual report. Sales and marketing appears as an aggregate operating expense, and GAAP does not require a per-customer breakdown. Investor-day materials may present "CAC payback" curves on selected cohorts, but those are not the same as an audited, per-head acquisition cost you can reconcile against the filing. Any single published CAC figure is reverse-engineered, not disclosed.
How many paying customers does DraftKings have?
The FY2024 annual report cites 3.5 million unique monthly payers. Important nuance: that is an activity-and-retention metric — people who paid in a given month — not a count of customers acquired during the year. Using it as a denominator for annual marketing spend would conflate retention with acquisition. Revenue per payer works out to roughly $1,363 a year, which is a yield figure, not an acquisition cost.
How did buying Jackpocket affect DraftKings' customer numbers?
DraftKings acquired Jackpocket for $750 million, with the deal verified at 30 June 2024. That is customer-base acquisition by M&A — buying an existing registered-user list and a lottery-courier distribution channel rather than bidding for installs through performance marketing. It distorts any blended acquisition-cost reading, because those users never passed through a marketing auction. A flat-looking blended CAC in a high-growth year is often explained by exactly this kind of inorganic purchase.
In how many US states is DraftKings sportsbook live?
DraftKings operates online sportsbook across 27 US states. Its first legal launch was New Jersey in August 2018, where it now holds roughly 27% market share per the New Jersey Division of Gaming Enforcement, verified December 2024. The 27-state spread matters for cost analysis because mature states like New Jersey carry low marginal acquisition cost, while newly competitive states carry high cost — and a single blended figure averages the two together.
Can I trust an "improving CAC payback" claim from the investor day?
Treat it as directional, not audited. The operator defines both the marketing costs counted and the customer denominator, and changing either moves the number materially. The claim also has to be read against the competitive year it covers — the ESPN Bet launch on 14 November 2023 raised auction prices across shared acquisition channels. An improving curve can coexist with rising true costs if the cohort or cost definitions are favourable.
What would change our conclusion?
We would reverse our position the moment DraftKings' FY2025 annual report discloses annual gross new customers alongside marketing spend, stated on the same basis as the investor-day payback cohorts — letting a reader divide one by the other and check the claim. Until the filed FY2025 10-K pairs those two figures with matching definitions, the reconciled CAC the query asks for cannot be audited from the public record, and the yield-not-cost distinction holds.