Push and pop are two formats many people start affiliate marketing with: a low entry barrier, huge volumes, a simple CPC or CPM auction. But cheapness has a price. In these networks you buy not only people but also bots, incentivized clicks, repeat impressions and zones that have never brought a single conversion. If you do not separate one from the other, your stats turn into noise, and your budget into payment for someone else's scripts. This guide covers how to filter push and pop traffic properly.
What push and pop traffic is
Push traffic consists of visits from browser notifications. At some point the user allowed a site to send messages, and now the network shows them ads in the form of such notifications. There is also in-page push — a banner styled as a notification that appears right on the publisher's page and works even without a subscription, including on iOS.
Pop traffic (popunder, popup) — the advertiser's page opens in a new window or tab when the visitor clicks anywhere on the publisher's site. A popunder usually opens "under" the current window, a popup on top of it.
| Parameter | Push / in-page push | Pop (popunder) |
|---|---|---|
| User action | Tap on the notification | Click anywhere on the site |
| Buying model | Mostly CPC | Mostly CPM |
| Volumes | Large | Very large |
| Typical problem | Bots, "subscription farms", list burnout | Instant closes, bots, repeat impressions |
| Where filtering matters | Weeding out junk zones and automated clicks | Weeding out bots and visits with no real view |
Both formats are sold in networks with thousands of publishers behind the volume: from quality media to sites that exist only to show ads. Which push and pop networks have source templates is in the traffic source catalog.
Why push and pop have so many bots
Push and pop are traffic from open placements. The network pays the publisher for impressions or clicks, which means the publisher has a motive to fake them. Hence several typical sources of junk:
- Automated clicks. Scripts that "tap" notifications or open popunders in emulators and headless browsers. Similar schemes with emulators and device farms are covered in detail in in-app traffic and fraud.
- Subscription farms. Push subscriber lists collected on devices nobody actually uses: servers, virtual machines, old phones in a rack.
- Data centers and proxies. Visits coming not from home and mobile networks but from hosting and cloud providers.
- Repeat impressions. The same person sees a popunder dozens of times a day, and every time it is a paid visit.
- Spy services and scanners. Services that collect other people's funnels monitor push and pop networks as closely as social networks: cheap traffic gets copied fastest.
More on how a bot differs from a person in bot traffic: signs and how to detect it.
Tip. Do not judge a network's quality by one campaign. Junk in push and pop is unevenly distributed: in a given network it is concentrated in a few dozen zones, and once you turn them off, the same network starts paying off.
Do you need a cloaker for push traffic
Traffic in push and pop networks is cheap, so "why pay for a filter on top" is a natural question. The answer: precisely because there are so many bots here, a filter in this traffic is about money.
- Do not show the offer to bots. Every bot that reaches the offer is a visit that ruins CR, EPC and your split test conclusions.
- Hide your funnel from spy services. If your landing page works well, it will be copied, and a week later there will be ten clones on the same traffic. How to fight that is covered in protection from spy services.
- See which placements are junk. A filter that explains its reasons shows where almost everything gets cut — that is your direct list of zones to turn off.
- Do not hurt your quality in the affiliate network's eyes. Bots that reach the offer produce fake leads and trash; the network sees a low approval rate and may cut your payout or turn off your traffic.
At the same time, a filter does not cancel the network's own rules: the ads and the page people see must comply with them, and for restricted verticals also with the law in the geo. That is covered in restricted ad categories.
If you are just figuring out how filtering works, start with what is cloaking: it walks through the click path step by step.
How to set up push and pop traffic filtering
Below is a practical setup that suits most push and pop campaigns.
Warm-up: the first clicks go to the White Page
The first visits after a campaign launch are noticeably dirtier than the rest: spy services monitoring fresh ads and automated visits from empty placements arrive at a new creative first. Warm-up sends the first N clicks of a flow to the White Page without checks, and those visits never see the offer. For native and push it makes sense to make warm-up longer than for search ads: these networks have more automated visits at the start.
Click limit per IP
On pop the same person can see your page many times, and bots often come from a limited set of addresses. A daily click limit per IP cuts both: after N visits from an address, the next ones see the White Page. For push and pop the limit is usually softer than for Google, because showing the page again to a real person is normal here.
Required JavaScript and a browser check
Simple bots and scrapers do not run JavaScript, or run it differently from a real browser. Requiring JS and blocking headless browsers cuts a large share of automated clicks without touching ordinary visitors: they always have JS.
Data center networks and visits with no ISP
Real push subscribers sit on home and mobile networks. Visits from hosting, cloud and networks with no visible ISP are a strong sign of a server. How such addresses are detected is covered in detail in VPN, proxy and data center IP detection.
Pass the zone ID in a parameter
The main tool for working with push and pop is a breakdown by placement. Almost every network can insert a zone or publisher macro into the link. Pass it into a campaign parameter:
https://your-landing.com/?sub1={zoneid}&sub2={campaignid}&cost={cost}
Macro names differ from network to network: {zoneid} in one, {zone}, {placement} or {site_id} in another. Check the specific network's documentation. More on parameters in UTM parameters and ad platform macros.
How it works in ArtisanClo
ArtisanClo has a ready platform preset for this format: "Native and push". You pick it as the traffic source in the first step of the flow, and protection in the second step is tuned for it automatically. Native specifics are covered in native advertising: Taboola, Outbrain, MGID.
| Parameter | Value for "Native and push" | Why |
|---|---|---|
| Block without ISP | On | Servers and farms often have no ISP |
| Warm-up | 100 clicks | Lots of automated visits at the start |
| Required JS | Yes | Simple bots trip over the script |
| Clicks per IP a day ("Balanced" level) | 5 | Native and push have many repeat impressions |
On top of that you choose the strictness: "Soft" cuts only IP blacklists and automated bots, "Balanced" adds VPN, proxies and data centers, "Strict" also adds IPv6, the live interaction check and a minimum time on page. For most push campaigns "Balanced" is a sensible starting point.
A few things that are especially useful for push and pop:
- A shared bot list. If an address has been confidently caught as a bot at any client of the service, its next visit to your flow gets the White Page right away. In open networks the same farms visit many advertisers, and this list works for you from the first click.
- A reason for every decision. In the click log every visit shows why it went to the White Page: "High-risk network", "Too many clicks from one IP", "Headless browser detected" and dozens of other reasons.
- Money by zone. If you pass the zone ID in
sub1, the "Money by slice" report shows clicks, conversions, revenue, spend and ROI for each zone. Turn off placements where the filter cuts almost everything in the network itself — that is how you stop paying for them. - Click cost from the link. The "From the link" cost model takes the price the network inserts via a macro, such as
cost={cost}, and spend is calculated for every click that reaches the offer. - A status code instead of a page. If filtered visits do not need to see any page, the White Page can be replaced with a bare 200, 403 or 404 response when you connect via the PHP file. A scraper or autoclicker simply has nothing to read, and a farm gets no content to learn from.
Connection works the same as for any other platform: a JS tag as the first line in the landing page's <head> or a PHP file on your server. Details in how to connect a cloaker to your site.
Tip. If your traffic goes through Keitaro or Binom, ArtisanClo can sit inside it as a filter or gateway — your tracker still handles routing, and the filter only answers whether to let the visit through.
How to read push and pop campaign stats
With push and pop it is easy to draw the wrong conclusion if you look at one number.
The pass rate on its own does not evaluate anything. It depends on the source: in one network most visits will reach the offer, in another fewer, and both can be normal. The warning sign is different: if almost nobody reaches the offer, the rules are most likely cutting real people along with the bots. In ArtisanClo the dashboard shows a hint in this case, and a notification warns you about the problem.
Look at a combination of metrics:
- CR and EPC by zone, not by the campaign as a whole;
- the distribution of filter reasons: if almost everything goes out on one reason, check whether that particular check is too strict;
- uniqueness: a high share of repeat visits on pop is a reason to lower the frequency cap in the network settings;
- ROI by day: push lists burn out, and a funnel that worked a week ago may stop paying off without any changes on your side. Many push and pop buyers send traffic to a smartlink instead of a single offer; it picks an offer for each visitor.
How to calculate profit correctly is covered in how to calculate profit and ROI.
Shadow mode before tightening
If you suspect the filter is cutting real people, do not turn it off blind. When connected via the PHP file there is shadow mode: for a few hours everyone goes to the offer, and the log marks whom the filter would have cut. If there are conversions among the "cut" ones, the rule is too strict. Keep shadow mode short: while it is on, the offer is open to everyone.
Common mistakes with push and pop
- One flow for all networks. Decision reasons and reports get mixed, and you cannot tell which network brings junk. One campaign — one flow.
- No zone ID in the link. Without it you see the average across the board and cannot turn off specific placements.
- Betting on "premium" without a filter. Expensive traffic has bots too, just a smaller share.
- Too strict an IP limit on pop. Repeat impressions to a real person are part of the format; a "1 click per IP" limit will cut them too.
- Conclusions from the first hundred clicks. Push and pop have high variance: wait for enough volume to make a decision.
- Installing the tag through a tag manager. The filter script must be the first line in
<head>with no async or defer, otherwise the page shows before the decision arrives.
If you are choosing a vertical for this traffic, first check whether the network allows it and whether licenses are needed in your geo — see restricted ad categories.
Summary
Push and pop traffic pays off when you pay for people, not for placements that fake impressions. That takes three things: a filter that weeds out bots, data centers and repeats before the offer; a zone ID in every link so you see money by placement; and the habit of regularly cleaning zones in the network itself. ArtisanClo covers the first two with a ready preset for native and push, decision reasons for every visit and reports by parameter. What the plans include is on the pricing page.
