How submissions are reviewed
Every invite submitted to useInvites goes through the same checks, in the same order, whoever submits it. Most get a decision within seconds. This page is the whole list — if a submission is turned down, it is because of something below, and you will be told which one.
What happens when you press submit
1. Automatic checks, at the door
Prohibited categories, link shorteners that hide where they go, contact details in the description, and a minimum length and word count. These are fixed rules, applied identically to everyone, and nothing gets past them.
2. A review of the listing itself
Whether the tip actually tells a reader something, whether the link belongs to the brand, whether the offer matches what the brand does. Most submissions get a decision in seconds.
3. A person, when it's genuinely unclear
Anything the automated review isn't confident about goes to a human — as does every brand that isn't on useInvites yet. Being sent to a person is not a rejection.
What the review checks
Each of these is judged on its own. The two marked must passmean the listing is wrong rather than weak — a link that goes somewhere else, or an offer that isn’t this brand’s. The rest send it to a person.
Pro tip substance
sent to a person if unclearContains at least one specific a reader can act on: a condition (e.g. 'make a card payment'), a timeframe (e.g. 'within 30 days'), a named step, a requirement, or a practical tip. Mentioning the reward alongside conditions is fine — the conditions themselves are the substance.
Nothing but vague encouragement ('great app, sign up!'). After stripping adjectives, no condition, step, or timeframe remains. Use 'unsure' if marginal.
Brand match
must passThe offer described plausibly belongs to this brand's actual product.
The offer clearly belongs to a different company or product than the brand it was filed under. Adjacent products from the same company are 'unsure', not 'fail'.
Link authenticity
must passThe link plausibly belongs to this brand's referral programme. Tracking parameters and referral subdomains are normal and are not a failure.
The link's domain is unrelated to the brand, or the destination is deliberately obscured.
Offer currency
sent to a person if unclearNo stated end date, or a stated end date still in the future.
States an end date that has already passed.
Language quality
sent to a person if unclearComprehensible to a general reader. Non-native but clear English passes — judge comprehensibility, never dialect or accent.
Machine-translated, incoherent, or garbled to the point that a reader cannot extract the meaning.
Substantive duplication
sent to a person if unclearSays something the brand's existing live tips do not already say.
Near-identical in substance to an existing live tip for the same brand, differing only in the code.
Brand reputability
sent to a person if unclearThe brand is a well-known, established company with a real referral programme. Examples: Uber, Monzo, Airbnb, DoorDash, Revolut, Cash App, Binance. If you recognise it as a real, operating company, it passes.
The brand name is unrecognisable, appears fictitious, is associated with known scams, or has no verifiable online presence. Niche but legitimate brands are 'unsure', not 'fail'.
The written description
At least 80 characters and 12 words, and — the part that actually matters — at least one specific thing a reader can act on. A useful test: strip out every adjective and every urgency phrase, and see whether anything is left.
Gets approved
“Upload a clear photo of your licence at step 2, or verification stalls for about 3 days. The bonus lands after your first 10 trips, not at sign-up.”
Gets turned down
“Use my link to get the best deal available today, amazing offer, sign up now and start saving straight away!”
If a decision looks wrong
Every turned-down submission appears in your dashboard with the reason attached, and two ways forward. If the problem is fixable, rewrite the description and resubmit — it goes straight back through review. If you think the decision itself was wrong, say so and a person reads it and replies.
An automated decision is never the last word.