Live chat reply times roughly double in the hour after a player takes a significant loss, and 49% of those chats never get a substantive answer at all. Those are the headline numbers from a support-benchmarking exercise that tracked 1,200 test accounts across 38 operators in the second half of 2024. The pattern held across licence jurisdictions and languages, which is the part worth sitting with.

What the benchmark actually measured

Testers opened a live chat session, stated a real problem (a voided bet, a stuck withdrawal, a bonus that didn't credit), and logged time to first human reply, time to resolution, and whether the session ended with an answer or a canned sign-off. Each operator was tested twice: once during a normal session with no recent loss activity, once within 60 minutes of the account recording a net loss above 40% of its session deposit.

Median time to first reply sat at 1 minute 51 seconds in the baseline condition. After a loss, the same operators averaged 3 minutes 44 seconds. Not catastrophic in isolation — but the tail is where it bites. The slowest decile went from 6 minutes to over 19, and 49% of post-loss sessions ended without a substantive answer, against 22% at baseline.

The "go quiet" number

That 49% figure covers three failure modes: chat closed by the operator, chat left open with no further agent message, and chats that ended in a template response that didn't address the question. The third was the most common by a wide margin — roughly six in ten of the quiet cases. Players often read a template as an answer and leave, which flatters the operator's resolution metrics while the actual issue goes nowhere.

Why the gap appears

Support leads I spoke to offered three explanations, only one of which is defensible.

The workload explanation is real. Losses cluster — a bad result on a Saturday afternoon or a major football fixture pushes contact volume up sharply and queue times follow. That's staffing, and it's solvable with better forecasting.

The second is escalation drag. Post-loss chats skew toward disputes: settled bets, cashout rules, bonus terms. Those take longer and get bounced to a second queue, which inflates first-reply time without anyone being slow on purpose.

The third is the uncomfortable one. A subset of operators appear to deprioritise chats from accounts showing distress signals — rapid re-deposits, chasing behaviour, repeated contacts within an hour. Some of this is automated routing; some is agents making a judgement call. Either way, slower replies to the players most likely to be in trouble is the opposite of what a responsible gambling framework is supposed to produce.

What the numbers don't tell you

Reply speed is a poor proxy for help quality. An operator answering in 30 seconds with a link to the terms page scores better than one taking four minutes to actually fix the problem. The benchmark tried to control for this with the substantive-answer flag, but that's a judgement call made by testers, not a clean metric.

Sample size also matters. 1,200 sessions across 38 operators is roughly 31 tests per operator — enough to spot a doubling in median reply time, not enough to rank operators against each other with confidence. Treat this as a directional finding about the industry, not a league table.

The question operators should be asking

If your queue triages on account value or recent activity, you already have the data to check whether loss events are pushing players down the queue. Most support stacks log both the loss timestamp and the chat timestamp. Joining them is a SQL query, not a project.

The open question is what operators do with the answer. Slower replies after a loss can be a staffing accident, a routing quirk, or a deliberate filter — and from the player's side of the chat window, all three look identical. If you're on the other end of a chat that's gone quiet after a rough session, the practical move is to screenshot the conversation and email it to the licensed entity's complaints address; it creates a timestamped record that a live chat window won't.