One comparison, three possible outcomes

Enforcement of a personal account is a matching problem rather than a declaration. An operator gathers a set of signals about each account, compares them, and runs the result against thresholds. This page explains the three outcomes, what each one means for the money, and why the middle outcome is the one that generates almost every real dispute.

Open the partner account Start with what the clause says

Joins the two: one artefact, one person expectedKeeps them apart: a verified fact that differsA person decides: an overlap a household also produces

Direct answerAn operator compares a fixed set of signals between accounts - identity fields, verified documents, contact details, payment instruments, device and network identifiers, and behaviour - and counts how many joining signals the two accounts share. Below the operator's threshold the pair is left alone; at or above it the outcome is predetermined under the terms; in between, a human decides. Only the first outcome is invisible, and only the last one can be argued with, which is why the middle band is where nearly every linked-account complaint begins.

The three outcomes

outcome 1

clear Left alone. The pair shares no joining signal, or shares one that the operator treats as weak on its own. Nothing happens, nobody is told, and the two accounts continue as unrelated relationships. Most accounts with a shared postal address live permanently in this outcome.

outcome 2

review A person decides. The signals are contradictory or the count sits near the threshold, so the pair goes to a human queue. This outcome is not a finding and not an accusation: it is the operator doing what the ambiguous signals on the table require. It is also the slowest, and the only one a reader can supply evidence to.

outcome 3

linked One customer, one account. The count is decisive, or the same payment instrument appears on both accounts, and the terms apply as written: a hold while the position is worked out, bets placed under the unrecognised account dealt with, and the balance resolved in favour of the account the operator treats as the real one.

Worked example - counting signals against a threshold (illustrative) Two rules run in sequence. First: a joining signal is decisive on its own when the name and date of birth also match. Second, for everything else: 3 or more joining signals is decisive, 1 or 2 goes to a human review, and 0 is clear. Pair X shares three joining signals - the same card, the same verified number, the same verified email - and the numbers match. Count: 3 - decisive. Pair Y shares two ambiguous signals, the same postal address and the same household network. Count of joining signals: 0 - clear, despite two overlaps. Pair Z shares one verified email but the two names differ. Count of joining signals: 1, and the first rule does not apply because the numbers do not match - human review. The arithmetic is the point: overlaps and joining signals are different counts, and the number that decides is the second one. A household can share four things and share no joining signal at all, which is exactly what pair Y illustrates.

Where the signals come from

What was declared at sign-up. Name, date of birth, address, email and telephone number. Cheap to compare, useful as a first filter, and weak on its own: a surname and an address are shared by families and by strangers who never met.
What was verified later. The document set that passed an identity check, and the payout destination that was verified. These carry the most separating weight, because a second passport and a second account in a different name are facts rather than artefacts - the identity desk covers how those documents are collected.
What the connection reveals. Device and browser characteristics, and the network the session arrived on. This is the layer that produces the most accidental matches, because devices and home networks are genuinely shared - and the layer a reader has least visibility into.
What behaviour adds. Timing, stake patterns, and whether the two accounts are ever active at the same moment. Weak as evidence and used as context, which is why it belongs to review rather than to the decisive count.

Why the middle outcome exists at all

Because a threshold that fired on any single shared signal would be wrong most of the time. Two adults in one house share a device; a worked example above shows that pair clearing entirely. A student sharing a flat shares an address and a network. A household shares a debit card when only one was ever issued. Each of those is ordinary, common, and produces overlaps without producing the thing the rule forbids.

So the operator puts the count in the middle of the range and sends it to a person, and the person is given a question rather than a verdict: is there a fact here - a document, a name, a payout account - that separates these two, or is every overlap explained by an address? That question is answerable, and the answer is usually available from the reader, which is what makes the appeal page the practical half of this desk.

The part nobody sees

An automatic clear produces no notification, no record a reader can see, and no visible consequence. That is not concealment: there is nothing to disclose, because nothing happened. The practical implication is worth stating once - a holder can never prove to themselves in advance that the comparison will go their way, because the comparison runs on data they cannot inspect. What a reader can do is what this desk is for: know which signals are joining, know which are theirs, and keep the honest explanation ready before it is asked for.

What this page is not. It reproduces no named operator's threshold, because the band differs by company, product and jurisdiction and moves with each terms revision. It also does not list what to keep off a shared account. The mechanism is described so a reader can read their own terms and their own situation; the arithmetic is shown so that nothing here has to be taken on trust.