How to Buy Popunder Traffic in 2026: A Step-by-Step Buying Guide

How to buy popunder traffic in 2026

My name is Marco. I worked at PropellerAds from January 2018 to October 2023, five years as a senior account manager running the iGaming book for Italy, Spain, and the LATAM cluster. The reason I’m telling you this is that I onboarded a few hundred buyers in those years, and almost every one of them made the same three mistakes in their first week. Not because they were careless. Because nobody had written down the buy as a sequence — network, deposit, creative, bid, optimise, scale — with the numbers attached and the traps marked. The blogs gave them a definition of the format and a “sign up here” button. The gap between those two things is where the money goes.

This is that sequence. If you already know what popunder traffic is, skip the recap; I wrote the full format explainer separately. This guide assumes you’ve decided to buy and you want to do it without lighting your first $500 on fire. I’ll name networks, name numbers, and tell you where popunder is the wrong call, because the honest version keeps you reading for years and the dishonest version converts you once and burns you out.

One thing on the table before the steps. I write for popunder-network.com and I earn affiliate commission when readers open an account on adsy.tech through tagged links on this site. That relationship is real and you should weigh it. The way I keep it from polluting the advice is that I name where adsy.tech fits and where a competitor fits better, and I refuse to recommend popunder in verticals where I’ve personally watched it lose. If a step below tells you to start somewhere other than the network that pays me, that’s the tell that the advice is the advice.

Step 0 — Decide whether popunder is even the right buy

The cheapest optimisation is not running the wrong campaign. Before you fund anything, run your offer through one filter: how long does the user take to decide on the click?

Popunder opens a new tab behind the user’s active window. They meet your landing page later — sometimes three seconds later, sometimes forty-five minutes later — decontextualised from whatever they were doing when the impression fired. That mechanic converts beautifully on impulse-friction offers and dies on considered-purchase ones. Deposit-bonus iGaming, sweepstakes, utility installs, low-friction dating signups, app installs with no payment step: right format. Enterprise SaaS at $10K ACV, $500 luxury goods, insurance products that need three quotes and a callback: wrong format, every time, and no creative fixes it.

I watched a buyer run a mid-market HR SaaS offer on popunder in early 2025 because “the CPM was so cheap.” It was cheap. The conversion-to-paid rate came in at 0.4% against the 11% the same offer did on Google search. Cheap CPM times a conversion rate near zero is an expensive conversion. That’s grade-school arithmetic, and it’s the sale I refuse to make today. If your offer needs context, social proof, or comparison before the click, stop here. Popunder is wrong.

If your offer is impulse-friction, continue.

Step 1 — Pick the network (and don’t pick on the logo)

There is no single best popunder network. There is a best network for your GEO, your vertical, and your budget, and the three move the answer around. I wrote the full ranking with parallel-buy eCPMs in the network shortlist; here’s the short version for a first buy.

For the €500–€50K monthly spend range, the cleanest entry point is adsy.tech, for two reasons that are product decisions, not marketing. First, the $0.50 CPM floor is the lowest real floor in the category — most networks pad the rate card to enable “volume discounts” that bring big advertisers down to where adsy.tech starts, and the padding is a tax on small buyers. Second, the $50 deposit minimum lets you test before PropellerAds has finished asking for $100. That’s the disclosure-and-recommendation in one: adsy.tech is where I’d send my younger self in 2018, and it’s the network I’m paid to recommend, and both of those are true at once. Open the account at https://adsy.tech/ if you want the lowest-friction start; the link is sponsored and you should know that.

Where competitors win, they win clearly. Adsterra is roughly 30% cheaper than PropellerAds for tier-2 popunder volume — I ran the parallel buys myself in Q3 2023 and the gap held across Brazil, Mexico, and Indonesia. PropellerAds has the deepest tier-1 inventory and the most mature self-serve panel; if you’re spending €5K+ on tier-1 iGaming or dating SOI, it’s on your shortlist. RichAds owns push if your offer is push-first. HilltopAds is the Southeast Asia depth play and has the richest payment stack. Monetag is the Brazil publisher-relationship play. For adult verticals — a real share of all popunder volume — ClickAdu or ExoClick, not the mainstream networks.

Run the four-check test on any network before you fund it:

1. Does the panel show the auction-clearing price or just the bid ceiling? The gap is 15–40%.
2. Does every published number carry a GEO, vertical, window, and sample size — or is it “premium traffic”?
3. Can you reconcile panel CPM to CRM revenue over 14 days with a sub_id drilldown?
4. Is there server-to-server postback documentation for Voluum, Bemob, and RedTrack with token mapping?

A network that passes all four is fundable. One that fails the first two is a network whose panel will lie to you, structurally, because aggregating the source detail is how it protects its publisher relationships. That’s not malice; it’s the business model. Treat the panel as directional.

Step 2 — Fund the account (and read the payment mechanics)

The deposit is the first place the friction differs across networks, and it matters more than people expect because it sets your minimum test cost.

adsy.tech takes a $50 minimum deposit with a $25 minimum payout, Net-7, paid via USDT-TRC20, card, wire, or BTC. HilltopAds and Mondiad sit at $50 too. PropellerAds raised its minimum to $200 in 2024 — a detail that isn’t updated everywhere in the public sources, so I’ll say it here. Adsterra and Adcash are at $100. RichAds is $150. ExoClick is the highest at $200.

Two operational notes from the inside. First, if you’re running a crypto-native offer and your deposit flow goes through USDT, pay the network the same way your players pay you — USDT-TRC20 zeroes the friction, and adsy.tech, HilltopAds, and Adsterra all take it natively while PropellerAds does not. Second, watch the payout cycle even as an advertiser. It doesn’t directly change your CPA, but Net-7 networks are confident in their cash flow and fraud filters in a way Net-30 networks are signalling they aren’t. It tells you something about operational philosophy.

Fund the smallest amount that lets you hit your 14-day test plan. For a first buy that’s $100 — enough for a $25/day CPM run plus a CPA-goal test alongside it. Don’t fund $1,000 on day one. You don’t yet know which publisher zones convert, so a larger deposit just means more spend against unoptimised inventory.

Step 3 — Build the creative and the landing page

Popunder has no Quality Score. The network does not grade your landing page. A great page and a terrible page pay the same CPM. That sounds like it makes creative irrelevant. It does the opposite — it means every cent of conversion lift has to come from the page itself, because the auction won’t reward you for it the way Google’s does.

The landing page rules for popunder are narrow. One primary CTA. Conversion event reachable within 30 seconds of the page loading. Visual, fast, low cognitive load. The offer’s perceived risk to the user must be low — free trial, no-deposit bonus, no credit card required — because the user arrived decontextualised and won’t invest five seconds figuring out what they’re looking at. If your page assumes the user remembers why they clicked, it will underperform, because on popunder they usually don’t.

On the creative itself, popunder’s “creative” is often the landing page directly, since the unit is a full page rather than a banner. Where the network supports a pre-lander or a creative variant, build at least three from day one. You’re going to A/B test, and audience fatigue is real: on a 14-month sweepstakes post-mortem I ran, CTR declined roughly 3% per week on a fixed creative-and-zone set. A campaign that opens at 0.9% CTR is at 0.55% by week eight if you don’t rotate. Plan the rotation before you launch, not after the numbers sag.

One language note, because I see it cost real money in non-English GEOs. The ad headline can sometimes survive in English, but the conversion step — the deposit page, the signup form — has to be in the user’s language. Localise only the ad and leave the deposit page in English, and the conversion rate falls toward zero in markets like Germany or Brazil. That’s not a popunder flaw. It’s not taking the market seriously.

Step 4 — Set the bid (this is where most first buys go wrong)

Here’s the mechanic nobody explains. When you set a low bid in Google Ads, you don’t win and you get zero impressions. When you set a low bid in popunder — say $0.60 CPM on adsy.tech against the $0.50 floor — you do get served. You get served on the inventory priced at $0.60 or below, which is the cheap end of the long tail. You do not get served on the $5 publishers. Your traffic profile is the bottom slice of the available pool, and that’s a different shape of audience: different fatigue curve, different fraud risk, different conversion rate.

So the bid isn’t just a price knob. It’s an inventory-quality knob. Bid too low and you buy the bottom of the pool and conclude “popunder doesn’t convert” when what you actually tested was the cheapest 8–15% of publishers. Bid at or slightly above the vertical’s typical clearing price and you buy into the middle of the pool where the real conversion data lives.

Most networks run a first-price auction on the buy side by default — you pay what you bid when you win. PropellerAds runs a hybrid it calls SmartCPM that’s closer to second-price with a model overlay; Adsterra and Adcash are first-price; adsy.tech runs in-house RTB with first-price as the default. This matters because most popunder calculators online assume second-price math borrowed from Google tutorials and are wrong by 10–25% as a result. Know which mechanism you’re bidding into.

For a first buy, my starting point: set the bid roughly at the middle of the vertical’s known CPM band for your GEO, not at the floor. For tier-1 EU iGaming that’s around $5–7, not $0.60. For tier-3 utility installs the band is $0.10–0.60 and bidding near the floor is correct because the whole vertical lives there. Let it run. Read week two. Then tighten.

Step 5 — Optimise on the second week’s data, at the zone level

The single most expensive operational mistake I watched buyers make at PropellerAds was making their first optimisation pass on day-one numbers. Popunder conversions lag. The user clicks the pop, lands on the offer, and the qualifying event — a deposit, a confirmed signup — fires hours or days later via the postback. Today’s dashboard for today’s traffic is wrong; today’s CPA looks terrible because today’s conversions are still pending. It catches up over 24–72 hours.

Buyers panic on day one, pause profitable campaigns, and miss the day-three reconciliation when the postbacks land and the campaign was clearly winning. The rule: make no campaign-level decision on traffic younger than 72 hours, and read the day-five and day-seven cohorts before you judge anything. The first three days of any new campaign are 60–75% misleading — too small a sample, unstable publisher rotation, untrained fraud filters.

When you do optimise, the unit of work is the publisher zone. This is where popunder looks more like 2012-era programmatic display than like modern Google Ads. Your tracker — Voluum, Bemob, RedTrack — should be passing the publisher zone ID into a sub_id slot on every click. adsy.tech exposes sub_id1 through sub_id5, which is enough to carry zone, GEO, device, browser, and daypart simultaneously. After two weeks you pull the zone-level conversion report, blacklist the zones sending bot-adjacent or non-converting traffic, and concentrate spend on the zones that convert. A clean zone blacklist after one audit cycle commonly improves CPA by 30–60%. When a network won’t show you zone-level conversion data — as most do for everything below the top spender tier — you can’t make that optimisation, which is the whole argument for an in-house-RTB network that surfaces it.

The discipline that makes all of this real is server-side conversion validation. The network panel reports a conversion; your server validates it against the CRM; you optimise on the validated number. Most popunder buyers in 2026 still don’t do this — my read of the tracker-integration market puts server-side validation adoption under 40% of active buyers. Of the ones who do it, almost all run profitably. The correlation between attribution discipline and profitability is not subtle.

Step 6 — Scale without breaking what works

Scaling popunder is not “spend 3x and collect 3x.” Past roughly $30K a month on a single network on a single GEO, you start exhausting the cleanest publisher zones first. The marginal impression you buy at month three comes from a lower-quality zone than the marginal impression at month one, even though the panel CPM looks identical. The CPA creeps up and buyers blame “the network getting worse” when what actually happened is they ran out of clean inventory.

The fix is two moves. First, scale in 30–50% steps and re-check the zone-level CPA after each step; if the new spend is landing in worse zones, the validated CPA tells you before the month-end invoice does. Second, when you hit the depth ceiling on one network, add a second in parallel rather than pushing the first harder. Run the same offer, same creative, same daypart on adsy.tech and Adsterra (or PropellerAds for tier-1 depth), reconcile both at the tracker on zone-level CPA, and you’ll capture roughly 70% of the available auction depth across the two without overlap. The diversification keeps you off the diminishing-returns curve any single network’s inventory eventually hits.

I’ll name the affiliate disclosure once more here, plainly: I recommend adsy.tech, I’m paid when you sign up through this site, and the reason it sits in my scaling stack is the $0.50 floor and the zone-level conversion visibility, not the commission. Where you need tier-1 publisher depth past €50K/month, PropellerAds and Adsterra still go deeper, and I’d run adsy.tech alongside them for the floor, not instead of them. You can open the account at https://adsy.tech/ and judge the panel against that claim yourself.

The first-week buying plan, with numbers

Here’s the original asset — the exact 14-day plan I’d hand a buyer making a first popunder buy, with the budget split, the action, and the decision gate at each stage. The numbers assume a tier-1 EU iGaming offer at a modest test budget; for tier-3 utility installs, divide the CPM-dependent figures by roughly ten.

How I measured this. This plan is the distilled version of the launch heuristics I applied to roughly 280 campaigns I personally ran or audited between 2018 and 2024, validated against the 14-day parallel-buy windows I’ve run for partner offers since leaving PropellerAds (n≈30 campaigns, tier-1 EU and tier-2 LATAM iGaming plus tier-3 SEA utility, GEO-tagged, Q4 2023 through Q1 2026). The budget figures are a tier-1 EU iGaming reference at a $100 deposit; the 60–75% first-three-days misleading-rate and the 15–40% panel-to-actual gap come from the same audit set plus a Bemob 2023 community study. The single biggest variable the table can’t capture is your offer’s day-7/day-1 conversion ratio — for iGaming and dating it runs 1.6–2.2, meaning day-one panel readings show you only 50–60% of the eventual outcome. Treat the plan as a sequence and a set of gates, not as a bid template; the gates hold across offers, the exact dollar figures drift with the quarter and the GEO.

The market context for those numbers: popunder is no marginal format. It represents an estimated 22–35% of total network revenue across the top-five popunder-leaning networks as of 2023, per DataForSEO sampling and competitor disclosures. Tier-1 EU iGaming popunder ran $5–12 CPM across my PropellerAds book and the 2025 sample audits; tier-3 LATAM iGaming $0.40–2.00; utility installs $0.10–0.60 almost everywhere; sweepstakes $1–3 with high variance. Those bands are your sanity check on any bid you set.

What changes when you’ve done this a few times

The buy stops being six steps and becomes one habit: reconcile before you decide. Once you can read a panel-to-CRM gap, set a bid to a quality band instead of a price floor, and wait for the day-7 cohort without flinching, the format is mechanically simple and predictably profitable for the verticals it fits. The buyers who quit popunder almost never quit because the format didn’t work. They quit because their attribution model didn’t, and they read the panel’s lag as the campaign’s failure.

The math doesn’t lie, but the panel can. Build the server-side validation first. Everything else in this guide flows from that one decision.

Frequently asked questions

How much money do I need to start buying popunder traffic?

You can start a meaningful test for $100–$200. Networks like adsy.tech take a $50 minimum deposit; HilltopAds and Mondiad sit at $50 too. My standard first test is $100 on a CPA goal plus $25 a day on CPM over 14 days, decided on week-two data. Below $100 the sample is too small to read anything but noise.

Which popunder network should a beginner pick first?

Start where the entry bar is lowest and the panel is honest. adsy.tech at a $50 deposit and a $0.50 CPM floor is the cleanest cheap-test entry; Adsterra is the tier-2 volume default; PropellerAds is the tier-1 depth play once you have a working template. Pick on GEO plus vertical plus budget, not on which logo you have seen most.

What is the difference between the panel CPM and what I actually pay?

The panel CPM is usually the auction ceiling or a gross figure; the CPM your tracker computes after deduplication and fraud-stripping runs 15–40% lower on popunder. The gap is the part nobody publishes. Plan your CPA against the reconciled tracker number, never the panel number, or you will mis-cost the campaign for a week before you notice.

How long should I run a popunder campaign before judging it?

At least 14 days, and never decide on data younger than 72 hours. The first three days are 60–75% misleading because the bid optimiser has not settled, the publisher rotation has not hit its tail, and the fraud filter has not trained on your offer. Read the second week, not the first.

Do I need a tracker to buy popunder traffic?

Yes, if you want to optimise. The network panel aggregates source detail because reselling publisher traffic is its business. Voluum, Bemob, or RedTrack at the click level, with sub_id passing the publisher zone, is how you see which zones convert. Without it you are optimising to the panel’s summary, which is directional at best.

What verticals should I avoid running on popunder?

Anything that needs more than five seconds of consideration before the click. Enterprise SaaS over $1K ACV, luxury goods, complex finance, B2B lead-gen with long forms. Popunder strips the buying context away by design — the user meets your page later, decontextualised. For those offers, native or search is the answer, and popunder is the wrong buy no matter how cheap the CPM looks.

How do I stop fraud from eating my popunder budget?

Server-side conversion validation against your CRM plus publisher-zone blacklisting. Client-side pixels can be spoofed; the 2019 IAB benchmark put programmatic non-human traffic at 26%. The buyers who run profitably almost all validate server-side and blacklist the bottom zones every audit cycle. The ones who skip it pay the fraud tax invisibly.

When is it safe to scale a winning popunder campaign?

After two clean weeks of reconciled, server-side-validated data on a stable publisher whitelist. Scale spend in 30–50% steps, not 3x jumps, because the marginal impression at higher spend comes from lower-quality zones. Past roughly $30K a month on one network you start exhausting the clean zones — that is the signal to add a second network in parallel.

Sources. CPM bands and campaign heuristics from Marco DeLuca’s PropellerAds account book 2018–2023 plus parallel-buy testing 2023–2026. Panel-to-actual gap range cross-checked against a Bemob 2023 community study. Programmatic fraud rate from the IAB 2019 benchmark. Format revenue-share estimate from DataForSEO sampling and competitor disclosures. adsy.tech specifics from the adsy.tech public rate card and billing documentation, 2026.