Quick answer: To vet an ad network before buying traffic, treat it as a supplier and make it prove four things before any volume flows: where the traffic comes from, that it passes sub-IDs you can act on, how it monitors and refunds invalid traffic, and what you owe and when. Then run a capped test and scale only on quality you have verified in your own data.
Networks and brokers can switch on an offer faster than you can recruit affiliates one by one. That speed is the reason to use them and the reason to vet them hard. A network you treat like a partner gets the benefit of the doubt. A network you treat like a supplier gets a scorecard, written terms and a cap until the numbers earn it more.
Why vet an ad network like a supplier, not a partner?
A network or traffic broker sits between you and the people who generate the clicks, leads and installs. When it works, one agreement switches on many affiliates who already run your vertical. When it does not, you pay for traffic you cannot see, from sources you did not approve, on terms you did not negotiate.
A partner relationship runs on trust. A supplier relationship runs on specifications, inspection and the right to reject what does not meet them. A new network belongs in the second category until it has a track record with you. That framing changes three things:
- You set the spec. Allowed traffic types, geos, creative rules and the payable event are yours to define, in writing, before launch.
- You inspect the goods. Sub-ID data, lead or install quality and downstream outcomes are checked against your own records, not the network's dashboard alone.
- You keep the right to reject. Scrubs, reversals and clawbacks for traffic that breaks the spec are agreed up front, not argued about after the invoice lands.
Good networks welcome this, because a buyer who verifies quality can scale. If you are still choosing a type of network, start with our guide on how to choose an affiliate network. This post assumes you have a candidate and need to decide whether to take its traffic.
How much of the traffic you buy is likely to be invalid?
More than most buyers assume, and the rate varies by channel and model. Every network faces the same bot and click-farm pressure as the rest of the market, so the vetting question is not whether it has invalid traffic but whether it finds it before you pay. Two current, at-scale measurements give the range:
- Lunio's 2026 Global Invalid Traffic Report, as reported by Advanced Television in January 2026, found an average invalid traffic rate of 8.51% across more than 2.7 billion paid ad clicks on major ad platforms between August 2024 and August 2025.
- The same Advanced Television coverage of the Lunio report says lead gen campaigns showed 32% higher invalid traffic rates than eCommerce, and gaming and iGaming carried the highest industry average at 18.49%.
- Fraudlogix's Q1 2026 ad fraud statistics measured an 18.12% invalid traffic rate across a sample of 26.3 billion ad impressions, with the United States at 20.37%.
The reports measure different things – paid platform clicks in one case, ad impressions in the other – so do not average them or treat either as your expected loss. The takeaway is directional: lead gen and gaming sit at the riskier end, so a network selling you lead gen or app traffic should be able to explain, specifically, how it filters it.
What should you ask an ad network before buying its traffic?
Use a scorecard so every candidate answers the same questions and you compare answers, not sales calls.
| Question to ask | What a good answer looks like | Red flag |
|---|---|---|
| Where does the traffic come from? | Named traffic types (search, social, native, email, push, content, rewarded) per source, disclosed before launch | "Our proprietary mix" or "a blend of premium sources" with no breakdown |
| Do you run your own media or resell other networks? | Clear split between owned media, direct affiliates and any sub-networks, with sub-networks disclosed | Vague answers about "partners"; no way to tell how many hops the traffic has taken |
| Will you pass a stable sub-ID per source? | Yes – a consistent, non-rotating source ID on every click and conversion, plus a sub-source field where useful | Sub-IDs blank, randomized or rotated so you cannot block or scale a single source |
| Is the traffic incentivized? | A plain yes or no per source, with the incentive described | Incentivized traffic mixed in undeclared, or "it depends" with no detail |
| How do you monitor fraud and invalid traffic? | Pre-bid or click-level filtering, post-conversion review, and a named process for scrubbing bad sources | "Our advertisers have never complained" in place of a process |
| How do you handle caps? | Hard daily and total caps enforced at the network, with overage unpaid | Caps treated as targets; overage billed "because it converted" |
| Can you isolate new sources? | Yes – a separate tracking link or offer ID for any new source so it can be judged on its own | New sources blended into a live campaign without notice |
One red flag on source disclosure, sub-IDs or caps is usually enough to stop. Those three control everything else: if you cannot see the source, you cannot block it, and if caps are soft, a bad source can spend your budget before you notice.
How do you check traffic-source transparency and sub-ID passthrough?
Ask for it, then test it. Transparency is the leverage point: a network only earns when its affiliates earn, and sources keep earning by giving the buyer enough information to trust them. A network that will not disclose where traffic comes from should not be running your offer – non-disclosure is not something to negotiate around with a lower payout.
- Read the sub-ID data, not the summary. Confirm every conversion arrives with a source ID that stays stable across days. A source whose ID keeps changing cannot be optimized or blocked.
- Match sources to the disclosed traffic types. If the network said "search and native," device, referrer and timing patterns should look like search and native.
- Separate incentivized traffic from incentivized creative. Incentivized traffic means the user is paid, in cash or virtual currency, for the action. Incentivized creative means the ad highlights the benefit a customer stands to gain, with nothing paid for engaging. Both can work, but downstream quality differs, and the network should tell you which you are getting.
Transparency cuts both ways. Tell the network what a good downstream outcome looks like – a funded account, a kept appointment, an install that reaches a key event – so it can steer affiliates toward traffic that closes, not just traffic that converts on the form.
Which affiliate networks allow your own lander and payment gateway?
Many networks let affiliates use their own landing or pre-landing pages after the advertiser approves them, but few legitimate networks let traffic sources route payments through their own gateway. As a buyer, require lander pre-approval in writing and keep checkout on your own payment processor, so refunds, chargebacks and customer data stay under your control.
- Own landers and pre-landers: acceptable with approval. Get a list of every page a source will use, review the claims on each and require re-approval for changes.
- Own payment gateway: treat a request to process your customers' payments outside your own checkout as a red flag. It separates the sale from your records and moves customer data somewhere you did not approve.
- Tracking: whichever pages are used, conversions should fire from your side, through server-side postback or your own pixel, not from a page the source controls.
If you are the affiliate asking, the answer depends on each advertiser's rules, so ask per offer rather than per network.
What contract and payment terms protect a traffic buyer?
Most network disputes are payment disputes in disguise. Settle the money questions in writing before launch. The terms worth putting on paper:
- The payable event, defined precisely. Not "a lead" but the exact fields, validations and criteria that make one billable – or, for apps, the specific install or post-install event.
- Scrub and reversal windows. How long you have to review conversions, what justifies a reversal (duplicates, fake data, undisclosed incentivization, out-of-geo, bot signatures) and how evidence is shared, with reversals netted against the next invoice.
- Clawbacks for confirmed fraud. If fraud is confirmed after payment, the network refunds or credits it. A network that will not agree to this is asking you to carry its sources' risk.
- Payment timing that matches your review window. Pay after the scrub window closes, not before. Prepayment to a new network or broker removes your only leverage if quality turns out to be bad.
- Caps that are binding. Daily and total caps, with conversions above cap unpaid unless you approved the overage in advance.
- Source and creative rules as contract terms. Allowed traffic types, prohibited methods, creative approval and geo limits belong in the agreement, with a stated consequence for breaking them.
- Data handling and consent. For lead gen, who captured consent, on which page, with what language, and how proof of consent is stored and shared on request.
Resistance to written scrub rules or binding caps tells you more about a candidate than its offer list does.
How should you run a test campaign with a new network?
Start capped, inspect everything and scale only on quality you have verified yourself. The first campaign is not about volume – it tests whether the scorecard answers hold up in live data.
- Launch on one offer with a hard daily cap. Pick an offer where downstream quality shows up fast, with a cap small enough that a bad week is cheap.
- Approve a short list of named sources. Do not let the network open every source on day one. Fewer sources make the sub-ID data readable.
- Check sub-ID integrity in the first days. Confirm every conversion carries a stable source ID and that the traffic patterns match what was disclosed.
- Judge on downstream outcomes, not the form fill or install. Compare each source against your trusted sources on the same offer.
- Run your scrub on schedule. Apply the agreed reversal rules and watch how the network responds – its handling of the first disputed batch predicts the rest of the relationship.
- Raise caps source by source. Increase volume for sources that pass, hold marginal ones and cut failing ones. Scaling on the blended average lets weak sources hide behind strong ones.
- Re-vet when anything changes. New sources, a new offer or a sudden jump in volume each restart the cycle at a cap.
If you are also deciding whether an unproven offer is worth testing at all, our guide on judging a new affiliate offer with no track record covers that side of the decision.
Aragon Premium works with advertisers on transactional, pay-on-performance terms. If you want to put this scorecard to a network directly, talk to our team.
FAQ
What is the difference between an ad network and a traffic broker? An ad network aggregates affiliates or publishers and manages tracking, payouts and offer approval for them. A traffic broker typically resells traffic it buys elsewhere, sometimes from other networks. Brokers add reach, but each extra hop hides the original source, so disclosure and sub-ID passthrough matter even more.
How do I know if a network's traffic is fraudulent? Look at sub-ID data, not totals. Warning signs include conversions clustered at odd hours, near-identical form-fill timing, narrow device or IP patterns, high duplicate rates and strong front-end conversion with weak downstream outcomes. Ask the network to explain any outlier source before you pay for it.
Should I prepay a new ad network? Avoid it. Prepayment removes your leverage if quality is poor, because reversals then mean chasing a refund rather than adjusting an invoice. Pay after your scrub window closes, with reversals netted against the next invoice. A network confident in its traffic should accept that until it has a track record with you.
What is sub-ID passthrough and why does it matter? Sub-ID passthrough means the network sends a consistent identifier for each traffic source with every click and conversion. It lets you block the sources that fail and scale the ones that work. Without it you can only judge the network as a whole.
Is incentivized traffic always bad? No. Incentivized traffic, where users are paid in cash or virtual currency to complete an action, can work well for the right offer, especially in rewarded and app categories. The problem is undisclosed incentivization. Require it declared per source, set the payable event deep enough to reflect real intent, and judge it on downstream quality.