Orders
A referral order is the moment your program turns a recommendation into revenue — a referee buys, the referrer earns, and the loop closes. Most orders need nothing from you: Bloop records the order, checks it, and issues the reward automatically. The Referral orders view exists for the small share that need a human decision — an order held for review, one flagged as suspicious, or a referral that happened off-platform and needs adding by hand.
Every order carries a status that decides whether the referrer's reward goes out. Understanding those statuses lets you pay genuine referrers fast while keeping a check on the rest. Referred customers are among the most valuable you will acquire — Wharton research on roughly 10,000 accounts found them about 18% less likely to churn and meaningfully more profitable over time (Schmitt, Skiera & Van den Bulte, Journal of Marketing, 2011). The referrer who brought them is exactly the person you cannot afford to annoy with a wrongly rejected reward.
Best practice: Make reviewing held orders a routine, not a fire drill. A flagged order sits on a 14-day hold and then follows your normal rules if you do nothing — so a quick weekly pass clears legitimate referrers before they wait too long and catches abuse before a reward is paid. Pair this view with the Fraud center for the cases that need closer evidence.
Order status workflow
Every referral order carries a referral status that controls whether the referrer reward is created. Three statuses appear in the view, plus one internal state that exists before a purchase happens:
Unused (internal)
The referee has claimed a discount but has not yet placed a referral order.
No — there is nothing to reward yet.
Pending
The order is recorded but the referrer reward is on hold, waiting for a review period or a fraud check to clear.
No — not until it approves.
Approved
The order passed all checks. The referrer reward has been (or will be) issued.
Yes — the reward is real and totals update.
Rejected
You declined the order. No referrer reward is issued, and the referrer is notified.
No — and it will not be.
The lifecycle runs in one direction. An order starts the moment a referred buyer checks out, lands in Pending, then resolves to exactly one terminal state:
Unused → Pending
A referee who claimed a discount places an order
A held order appears; nothing paid yet
Pending → Approved (automatic)
The review period or 14-day fraud hold ends with no rejection
Reward is created; totals tick up
Pending → Approved (manual)
You approve the order yourself
Same as above, but immediately
Pending → Rejected (manual)
You reject the order
No reward; referrer gets a "referral denied" notice
The status is the gate on your budget. An order in Pending has cost you nothing yet — the referrer reward has not been created — so the time an order spends pending is your only cheap window to catch a problem before money leaves your account. The moment it flips to Approved, the reward is real and the referrer's totals update.
Approval is no longer the point of no return, though. If a referred order is later cancelled or fully refunded, Bloop can revoke the reward it already issued — deleting an unused discount code, debiting store credit back, or reversing an unpaid cash or unfulfilled custom reward. Which events trigger this is set by the revoke trigger in Order settings; see reward revocation. Your review period is still the cleaner guard — catching a bad order before the reward exists beats clawing one back afterward — but a genuine cancellation or refund after approval no longer slips through.
Decide quickly: which orders actually need you
Most referral orders never reach this view, and of the ones that do, only a fraction need real scrutiny. The skill is triage — knowing at a glance which holds to act on and which to leave alone.
Pending, no fraud flag, review period running
A normal order waiting out your configured delay
Nothing — it approves itself when the timer ends. Only act if you want to pay the referrer sooner.
Pending, fraud-flagged, 14-day hold
Bloop saw a suspicious pattern — review it in the Fraud center
Check the evidence and approve or reject before the 14 days run out.
Pending, unusually high-value order, no flag
A legitimate big order, or worth a quick sanity check
Approve to pay faster, or glance at the referrer/referee pair if the value is odd for your store.
Pending, referee returned the order during the window
A cancellation or refund before approval
Reject so you don't pay for a sale that no longer exists.
A referral that never appeared
A phone order, or a code Bloop couldn't capture
Add it manually so the right referrer gets credited.
Act on the holds, ignore the rest. An unflagged pending order needs nothing from you — leaving it alone produces the same outcome as approving it, just a little later. Spend your attention on fraud-flagged holds, where letting the reward pay out is the expensive mistake. For the wider stance on catching abuse without slowing genuine customers, see preventing referral fraud without blocking real customers.
When an order is held
A referral order lands in Pending for one of two reasons:
Reward review period. If your campaign sets a delay before rewards approve, the order waits that many hours. A delay of
0approves instantly. See Create and configure a referral campaign.14-day fraud hold. If Bloop flags the order as suspicious, it is placed on a mandatory 14-day hold so you can review it before any reward goes out. This overrides your review period.
When the period or hold ends without a rejection, the referrer reward is created. A fraud-flagged order always waits the full 14 days regardless of your review period. An unflagged order waits only your configured review period.
How long a review period should be
There is a genuine trade-off. A longer delay gives you more time to spot a return or chargeback before paying the referrer — but a slow reward is a weaker reason to share again. A few patterns that tend to work:
A short delay (a day or two) suits most stores. It covers the common case — an immediate cancellation or an obvious mistake — without making honest referrers wait. The fraud hold already gives you 14 days on suspicious orders.
A longer delay makes sense if your return rate is high or order values are large. Match the delay to your real refund timeline.
A zero delay suits low-risk, low-AOV stores that want maximum momentum. Instant rewards are the strongest nudge to share again; if your fraud exposure is small, the speed may be worth more than the safety margin.
Approve or reject an order
For any pending order:
Open the Referral orders view and find the order.
Review the referee, the referrer, the order value, and any fraud flags.
Approve to issue the referrer reward, or Reject to decline it.
Approving a flagged order clears its fraud status and closes the related case. Rejecting a flagged order marks it disqualified, records the decision in the fraud history, and sends the referrer a "referral denied" notification. Suspicious orders are also reviewable in the Fraud center.
When to step in manually — and when not to
Manual approval and manual rejection solve different problems, and most days you should use neither:
A clean, unflagged order you happen to be looking at
Leave it — let the timer approve it
Acting manually changes nothing except making you do work
A genuine referrer is waiting and you want to thank them sooner
Approve early
Speed makes the reward a stronger reason to share again
A fraud-flagged order with weak evidence (shared office IP, family)
Approve
A real advocate is worth far more than one reward; don't punish ambiguity
A fraud-flagged order with a clear, repeated abuse pattern
Reject (and consider a ban in the Fraud center)
This is the case the hold exists for
The referee cancelled or refunded before the hold ended
Reject
There is no genuine sale to reward
When you are unsure, weigh the cost of each wrong call. Rejecting a genuine order annoys a real advocate who may never share again. Approving a fraudulent one costs you one reward. When evidence is genuinely ambiguous, approving is usually the gentler default. The exception is a clear, repeated pattern — that belongs in the Fraud center, where you can ban rather than wave through.
Manually add a referral order
Sometimes a referral happens outside the normal flow — a customer ordered by phone, or used a code Bloop could not capture. You can add the order by hand:
In the Referral orders view, choose to add an order.
Enter the Shopify order details — order number, amount, and the customer (by email or customer ID).
Bloop validates the order before saving:
If the order is already linked to a referral, you cannot add it again.
If the customer is already tied to a different order, Bloop blocks the duplicate so a referrer cannot be credited twice.
Choose the referrer to credit. If the buyer already claimed a referral discount, Bloop locks the order to the referrer who referred them, so you credit the right person.
Pick the campaign to credit the order to. The dropdown lists only active campaigns — paused or inactive ones can't be selected.
Optionally choose to issue the referrer reward immediately when you save.
A manually added order is normally recorded as Approved and increases the referrer's successful-referral count, total referees, and total revenue. The exception is a custom reward: because a custom reward is fulfilled by hand, the order is added as Pending and waits for you to approve it — it shows a plain "Pending" label with no auto-approve countdown, rather than issuing on its own.
A worked example
The values below are illustrative — they show how to reason about a held order, not measured Bloop results.
A customer service rep tells you a long-time customer, Maya, placed a $480 order over the phone and mentioned a friend had recommended you. There is no referral code on the order because it never went through your storefront. Here is the loop:
Confirm the referrer exists. You check that the friend Maya named is enrolled in the program. They are, so there is a real referrer to credit.
Add the order manually. In the Referral orders view you enter the Shopify order number, the $480 amount, and Maya as the customer. Bloop checks for duplicates — Maya is not tied to another referral order, so it saves cleanly.
Credit the right person. You select the friend as the referrer. The order saves as Approved and their successful-referral count, total referees, and total revenue all tick up.
Decide on the reward timing. You issue the referrer reward immediately — this is a verified, off-platform referral with no fraud risk.
A second example: a held order you should slow down on
Now contrast a held order you should not rush. A new account — created the same week — refers three orders in two days, all shipping to the same address, and Bloop flags them on a 14-day hold:
Read the flag, not your gut. Three orders sharing an address from a brand-new referrer account is the textbook self-referral shape.
Check the cheap-vs-expensive math. Leaving the orders held costs you nothing — the rewards have not been created. Approving on sight would create three rewards for what is likely one person rewarding themselves.
Take it to the Fraud center. Because this is a pattern across several orders from one referrer, investigate it in the Fraud center rather than deciding order-by-order. If the evidence holds up, reject the referrals and ban the referrer; if it turns out to be three genuine friends at a shared address, approve.
The contrast is the whole lesson: with Maya, speed was free and kindness was cheap, so you paid fast. With the three same-address orders, patience was free and haste was expensive, so you waited.
Common mistakes to avoid
Treating every pending order as a task. Unflagged orders approve themselves. Chasing them wastes the attention you need for the flagged ones.
Ignoring the held queue for weeks. A fraud flag only protects you if you review it inside the 14-day window. Past that, the hold expires and the reward follows your normal rules.
Rejecting a genuine order to be safe. A real advocate denied a reward they earned often stops sharing. When evidence is ambiguous, the gentle default usually costs less than the lost referrer.
Setting the review delay by gut, not by your refund timeline. Too short and you pay before returns settle; too long and honest referrers wait for no reason.
Adding a manual order without confirming the referrer. Crediting the wrong person corrupts your numbers and hands out a reward to no one's benefit.
Relying on revocation instead of your review delay. Bloop can revoke a reward when an order is cancelled or fully refunded (see the revoke trigger), but a reward the referrer has already used can't be clawed back. A sensible review delay still catches bad orders before the reward is ever created — the cleaner guard.
Edge cases
A refund after approval. Approving issues the reward. If the order is later cancelled or fully refunded, Bloop revokes the reward according to your revoke trigger — unless the referrer already used it, which can't be reversed. A partial refund never revokes. For return-heavy catalogues, a longer review delay is still the cleaner guard than issuing fast and revoking later.
A cancellation during the hold. If the referee cancels or is refunded while the order is still Pending, reject it — there is no completed sale to reward, and rejecting before the timer fires stops the payout before it is ever created.
A duplicate the validator blocks. If you try to add an order already linked to a referral, or a customer already tied to another order, Bloop refuses the save. That is the safeguard against double-crediting working — not an error to override.
A buyer who already claimed a referee discount. Bloop locks a manual order to the referrer who originally referred that buyer, so you cannot accidentally credit a different referrer for the same customer.
A flagged order you approve. Approving clears the fraud status and closes the case. If you were wrong, the fraud history still records that you approved it.
Next steps
Investigate flagged orders in the Fraud center.
Learn how referrer rewards are calculated in Referrer and referee rewards.
See the broader stance in preventing referral fraud without blocking real customers.
Understand the wider problem in How to prevent referral fraud.
Last updated