KYC Document Upload Rejects 43% of First Tries at Night
A 43% first-attempt rejection rate on KYC document uploads between 10 p.m. and 6 a.m. local time is the headline figure in a review of 1.2 million verification submissions processed across 14 US-facing operators between January and September 2024. The same dataset shows a 21% first-try rejection rate during daytime hours, meaning a player who signs up after dinner is roughly twice as likely to have their ID bounced before they ever reach a cashier. The gap holds after controlling for document type, device, and operator, which is why compliance teams are now looking at the clock as a risk variable rather than a footnote.
The 43% figure and where it comes from
The numbers come from a verification vendor that handles identity checks for a mix of sportsbook, casino, and poker brands licensed in states including New Jersey, Pennsylvania, Michigan, and Colorado. The dataset covers 1,204,880 upload attempts from roughly 640,000 distinct accounts. Each attempt is a single submission of a document image — a driver's license, passport, utility bill, or selfie — and a rejection means the automated system or a human reviewer returned it for a redo.
The hour-by-hour breakdown is where it gets interesting. Rejections climb steadily after 8 p.m., peak between 11 p.m. and 2 a.m. at just under 47%, and fall back toward the daytime baseline by 7 a.m. The 43% figure in the headline is the average across the full overnight window, not the peak.
Two caveats matter for anyone citing this. First, it is a single vendor's book of business, not a census of the US market; operators using other vendors will have different absolute rates, though the overnight pattern is consistent enough across the sample that it is unlikely to be a vendor artifact. Second, "rejection" here includes soft failures — blurry images, glare, cropped corners — not just suspected fraud. That distinction is the whole story.
What counts as a rejection
The vendor's taxonomy breaks rejections into four buckets: image quality (glare, blur, low light, partial document), mismatch (name or address doesn't match the account), expiry (document out of date), and authenticity flags (possible tampering or synthetic ID). Image quality accounts for 61% of overnight rejections versus 44% during the day. Authenticity flags barely move — 3.1% overnight versus 2.8% daytime. In other words, the night spike is mostly people taking bad photos, not a surge of fraudsters.
Why the clock changes the outcome
Three explanations get tossed around in compliance circles, and the data supports all three to varying degrees.
Lighting and device conditions
The single largest driver is photographic. A driver's license photographed under a 60-watt bulb at 1 a.m. produces glare on the laminate, washed-out text, and shadows across the date of birth. The vendor's own image-quality scoring shows a 19-point drop in average sharpness scores for uploads between midnight and 4 a.m. compared with midday. Phone cameras are also more likely to be in low-light mode, which introduces noise and motion blur that the OCR engine reads as a mismatch.
The player is tired, rushed, or both
Behavioral data from the same vendor shows overnight uploads are 34% more likely to be completed in under 15 seconds from the moment the upload screen loads. Fast submissions correlate with fewer retakes before hitting send. A player who has just deposited, or is trying to deposit, and hits a document wall at 1 a.m. is not carefully framing the shot. They are holding the license at arm's length and hoping.
Staffing and queue times
Not all rejections are automated. Operators with manual review queues staff them thinner overnight. The vendor's median time-to-decision on manual reviews stretches from 4.2 minutes during business hours to 18.7 minutes overnight, and the share of submissions routed to manual review rises from 12% to 23% because automated confidence scores are lower on poor images. A longer queue doesn't cause a rejection by itself, but it does mean a player who fails once at 1 a.m. may wait 20 minutes for the verdict and then abandon the process entirely.
What a rejected upload actually costs the operator
The player-facing annoyance is obvious. The operator-facing math is less so, and it is the reason this is becoming a board-level conversation at some brands.
The vendor tracked what happened to accounts after a first-attempt rejection. Of players rejected once overnight, 38% never completed verification within seven days. During the day, that figure is 24%. Of those who never completed, 71% had made at least one deposit attempt — meaning the operator took the money, or tried to, and then could not let the player withdraw or, in some cases, bet.
That is a regulatory exposure, not just a customer-service one. Under the AML program requirements that apply to licensed operators, a customer who cannot be verified within a reasonable window faces restrictions on activity. Several state regulators have fined operators for letting unverified players continue to wager, and the overnight rejection pattern is exactly the kind of gap that shows up in a routine audit. A 43% first-try failure rate concentrated in a six-to-eight-hour window is a lot of players sitting in a verification limbo that compliance departments are supposed to be closing, not creating.
There is also the simpler cost: acquisition. A sportsbook that pays $180 to $320 per funded account in affiliate and marketing spend, depending on state and channel, is watching a meaningful share of that spend evaporate because a photo was blurry. If 38% of overnight rejections never convert, and overnight signups are roughly 31% of daily volume for the operators in the sample, the arithmetic gets uncomfortable fast.
The re-upload loop
Most operators allow three to five re-upload attempts before escalating to manual review or locking the account. The vendor data shows the average overnight rejector needs 2.4 attempts to succeed, versus 1.6 during the day. That means the effective cost per verified player rises with every retry — in reviewer time, in SMS and email notifications, and in the not-insignificant number of players who give up at attempt two.
What operators are actually changing
The fixes fall into three categories, and none of them are exotic.
Capture guidance at the point of upload. Operators that added a live camera capture — forcing the phone's camera rather than allowing a gallery upload — saw overnight rejection rates drop by 11 to 14 percentage points in A/B tests run in the third quarter. The camera can detect glare and blur in real time and prompt a retake before submission. One mid-size casino brand reported its overnight first-try rejection rate falling from 41% to 29% after a live-capture rollout in July.
Timing the ask. Some operators now delay the KYC prompt for players who sign up late at night, letting them deposit and play under a provisional limit while nudging them to verify the next morning with a reminder. This is legally sensitive — the provisional window has to be short and the player has to be restricted from withdrawing until verified — but it moves the document capture to daylight, when the rejection rate is half as high.
Staffing the queue to match the demand. A few brands have shifted manual review capacity to cover the 10 p.m. to 2 a.m. window, which is when both signups and rejections peak. The vendor's data suggests the overnight manual queue is understaffed relative to volume by a factor of roughly 1.8.
None of these are novel. What is new is treating the clock as a variable worth measuring. For years, KYC rejection rates were reported as a single blended number, which hid the overnight spike entirely. An operator reporting a 28% blended rejection rate might have a 21% daytime rate and a 43% overnight rate and never know it.
What the regulators might ask next
State gaming regulators have not, to date, published guidance specific to time-of-day verification performance. But the audit trail exists, and the question of whether an operator "reasonably" verified a customer is easier to answer when the operator can show it measured its own rejection rates and acted on them. A 43% overnight first-try failure rate is not a violation on its own. It becomes one if it persists for years and nobody looked.
The open question is whether the industry will treat this as an engineering problem — better camera capture, better lighting prompts, better queue staffing — or as a customer-experience problem to be papered over with more retry attempts. The former reduces the number of unverified players sitting on deposits. The latter just moves them to a different queue. Either way, the next time someone quotes a single KYC rejection rate for an operator, the first follow-up should be: at what hour, and measured how? Players who find themselves photographing a license by lamplight at 1 a.m. might reasonably ask the same thing — and, if the retake fails twice, consider waiting until morning before trying again.