The useful distinction is not "good source versus bad source". It is whether the operator can maintain source integrity from placement to settled revenue.
The failure mode is familiar: the approved property is clean, the postback chain is nominally correct, and the first cohort produces a healthy-looking FTD and GGR curve. Two reconciliation windows later, the mix is different: resold inventory, unapproved sub-affiliates, brand-bid leakage, incentive traffic or synthetic sessions, and the original approval no longer describes the traffic being paid for.
Quality is therefore an evidence problem. The manager needs placement-level lineage, stable click identifiers, cohort economics and enough first-party telemetry to tell an acquisition change from an attribution artefact. The channel label is merely an index.
The live sources that still matter
There is no single best source. The right source depends on the market, licence, offer and the kind of customer the operator can retain. In practice, managers keep seeing the same live channels:
- SEO and editorial comparison pages. Slower to build, but the page, query, placement and audience can be reviewed before launch.
- Paid search. Fast feedback and strong intent, but certification, brand-bid rules and market eligibility must be checked for every target GEO.
- Creators, streamers and communities. Useful when the creator has a real, identifiable audience and the commercial relationship is disclosed.
- Telegram, email and push. Direct distribution can work well when consent, list origin and unsubscribe controls are visible.
- Native, display and programmatic. Scalable, but placement lists, viewability, fraud controls and postback quality matter more than the headline volume.
- Sub-affiliate networks. Potentially large, but the operator must know who is actually buying or publishing the traffic underneath the network account.
The practical answer to "where should we buy traffic?" is simple: start where the source can show a real placement, a real audience and a repeatable conversion path. A channel name is not proof of quality.
What traffic substitution looks like
Traffic substitution is when the source presented for approval is not the source that ultimately supplies the user. It can be an innocent reseller with poor visibility, or a deliberate switch after approval: one clean site is shown to the manager while cheaper pop, incentivised, bot or brand-bidding traffic is sent once the link is live.
Typical signals include:
- the approved URL and the referrer pattern do not match;
- a placement suddenly redirects through a new domain or tracker;
- the GEO, language, device mix or time-of-day distribution changes without a campaign change;
- the partner cannot provide a placement list or sub-affiliate names;
- click volume rises while session depth, verification or retention falls;
- a large share of conversions arrives with repeated device, IP or click patterns;
- the partner reports a source that the operator cannot reproduce in its own logs.
Do not call one signal fraud by itself. Treat it as a reason to pause, preserve the logs and compare the partner report with first-party data. The objective is to find the substitution before commissions and bonus costs make the source look profitable.
Read GGR as a cohort signal, not a payout base
GGR is useful at the top of the funnel economics, but a first-week GGR spike is not evidence of durable value and is not automatically the affiliate payout base. The reconciliation should follow the programme's defined NGR waterfall and the hold window, with the same cohort key used for acquisition, fraud and finance.
Read GGR together with:
- bonus cost and bonus-abuse rate;
- chargebacks, failed payments and fraud adjustments;
- KYC pass rate and restricted accounts;
- retention after 7, 30 and 90 days;
- NGR or the programme's defined payable revenue;
- complaints, self-exclusions and market-compliance events.
A source can generate impressive first-week GGR while producing weak NGR after bonus costs, reversals and payment friction. Conversely, a smaller SEO or community source may have a lower FTD spike but stronger retained value. Compare sources only after fixing the cohort window, revenue waterfall, attribution rule and hold period.
The short checklist
Before approving a partner, answer these six questions:
- Who owns the source? Record the entity, person, domain, channel and any sub-affiliates.
- Where will the traffic come from? Ask for live URLs, profiles, communities and paid platforms, not only a dashboard screenshot.
- Which GEO and audience will see the offer? Compare the declared markets with analytics, language and customer location.
- What exactly will the user see? Review the creative, landing page, terms and disclosure before launch.
- Can the conversion be reproduced? Test click IDs, sub-IDs, postbacks, duplicate events and reversals end to end.
- What happens when the source breaks a rule? The agreement needs a pause, audit and removal process.
If a partner cannot answer one of these questions, more volume is not the answer. The missing evidence is the risk.
What quality looks like in the data
Do not score a source on registrations alone. A useful report includes:
| Signal | What it tells the manager |
|---|---|
| Verified registrations | Whether the event represents a real eligible user |
| First-time depositors | Whether acquisition reached the intended action |
| Duplicate and invalid rate | Whether the source sends repeat or manipulated users |
| Chargebacks and reversals | Whether the first conversion survives review |
| Retention by cohort | Whether the source creates customers rather than one-day events |
| Complaints and exclusions | Whether the audience or message creates regulatory risk |
Compare cohorts by partner, placement and GEO. A source with fewer registrations but stronger retention can be more valuable than a source that wins a volume table for one week.
Check the source before the numbers
Ask each partner for a source map: URL or channel, format, target GEO, audience, traffic method and owner. For SEO, inspect the actual page and its outbound links. For Telegram or social traffic, review the channel history and disclosure. For paid traffic, record the platform, account owner and certification status where the market requires it. For sub-affiliates, require names and approval before launch.
Google's gambling advertising policy makes the destination check explicit: affiliate and comparison sites must link only to gambling entities authorised for the country being targeted, and the relationship between advertiser, gambling entity and domain must be clear. A manager cannot outsource that check to the tracker.
Review the landing page like a customer
Open the approved URL in the target GEO and on a phone. Check that the page says:
- who operates the product and where it is available;
- who is eligible and what the offer requires;
- how to contact support;
- which responsible-use or age information applies;
- that the commercial relationship is disclosed where it affects trust.
Reject pages that hide a material condition, impersonate an independent customer, use a misleading brand name or redirect through an unknown domain. Measurement links are normal. A redirect that conceals the final product or changes the offer is not a quality-control shortcut.
The FTC's endorsement guidance says a material relationship should be clear and conspicuous where the audience sees the recommendation, not buried in a profile or footer.
Keep the event lineage reconstructable
Agree on the parameter contract before launch and document:
- partner ID and placement ID;
- click ID and sub-ID format;
- conversion events, states and their definitions;
- postback URL, allowed macros and signing/authentication;
- deduplication key and replay handling;
- reversal, hold and payment windows;
- the join key used by tracker, PAM/CRM and finance.
Run a synthetic click through the real redirect chain and reconcile the click, registration, FTD, settlement and reversal records across systems. Confirm that the click ID survives every hop, the postback is idempotent, late events are handled, and a reversal leaves both the payable report and the partner export. A green test pixel is not an integration test. Our postback troubleshooting guide covers macro drift, encoded parameters, duplicate callbacks and event-name mismatches.
Red flags that deserve a pause
Pause a source while it is investigated when:
- the partner cannot identify the page or channel that sent the traffic;
- the approved URL changes to an unknown redirect;
- brand terms appear in paid search without permission;
- a placement removes an important condition or disclosure;
- registrations rise but verification, deposits or retention do not;
- a new sub-affiliate appears without approval;
- complaints arrive from a market the partner said it did not target.
Keep the URL, creative, click IDs, timestamps and event log so the review can be repeated. A pause is a control, not an accusation.
How ANTIMEDIA works with affiliate teams
ANTIMEDIA is built around the part of affiliate traffic that is easiest to lose in a volume report: context.
We publish editorial pages for organic reach, search visibility and discovery. A partner placement should be understandable on its own: the reader can see what the article is about, what is being linked and why the source is relevant. We do not build traffic on cloaking, fake testimonials or hidden redirects.
For an affiliate team, that means a placement can be reviewed before launch, linked to a specific article and measured against a defined URL rather than a vague promise of exposure. The source is named, the offer is honest and the conversion path is testable.
Approval scorecard
| Check | Pass when | Hold when |
|---|---|---|
| Ownership | Person, entity and source are documented | Account uses unnamed sub-affiliates |
| Source | Live placement and traffic method are visible | Only screenshots or estimates |
| GEO | Audience and eligibility match | Country is unclear or inconsistent |
| Creative | Conditions and relationship are clear | Claims or terms are hidden |
| Tracking | Test event and postback reconcile | IDs or macros are unverified |
| Quality | Retention and reversals are measurable | Payment relies on raw registrations |
Approve when every row passes. If one row is on hold, ask for evidence before raising the commission or opening more GEOs.
The decision rule
Quality affiliate traffic is not the biggest first-day number. It is traffic a manager can explain: who sent it, where the user saw the offer, which market applied, how the event was tracked and what happened after conversion.
That is slower than buying anonymous volume. It is also the only model that leaves the programme with a trustworthy source, a defensible report and a partner worth scaling.