Popunder vs Google Ads: which actually converts for impulse offers

My name is Marco. I spent five years at PropellerAds, running the tier-1 iGaming book from 2018 to October 2023. The reason I am telling you this up front is that everything below is going to compare a format the industry calls intrusive against the format the industry calls premium, and you should know which side of the desk I sat on.

The short answer is: popunder converts for impulse-friction offers and dies on considered-purchase offers. Google Ads is the inverse. The CPM math tells the story, but only if you know which line of the math the panel is showing you.

Why this question keeps getting asked badly

Most “popunder vs Google Ads” pages on the internet either skip the math, or quote rate-card CPMs against tCPA averages and pretend that’s a comparison. It is not. Popunder runs on CPM. Google Ads runs on CPC with Quality Score multipliers and tCPA bidding overlays. Comparing $1.80 CPM popunder against $2.10 CPC Google Ads is not a comparison — it is two different units.

The honest comparison is cost per converted user, by vertical, by GEO, on a campaign that ran long enough to stabilise. I have those numbers from my PropellerAds period and from buyer-side testing after I left. I’ll give them to you below.

The format mechanics, briefly

A popunder opens a new browser tab or window behind the user’s current tab. They see it when they switch tabs or close the original one. The unit is a full landing page, not a 728x90 banner inside a content slot. The user has already left the publisher when they see your offer, which means there is no in-stream context fighting your offer for attention.

A Google Ad runs in three main places: search results, the Display Network, and YouTube. Search Ads sit above organic results when the user has already expressed intent (“buy X near me”). Display ads sit inside content slots, competing with whatever the user came to read. YouTube ads interrupt video.

The two formats are not substitutes. They are different layers of the funnel. The mistake most affiliate buyers make is treating them as substitutes because both are “buy traffic.” They aren’t.

CPM and CPC, head to head

Here are numbers I’ll defend. Sources at the bottom of the post.

The first thing to notice: popunder’s cost advantage is real, and it widens at the bottom of the funnel where intent is low and impressions are cheap. The second thing to notice: Google’s CPC advantage on enterprise SaaS isn’t a flaw in popunder, it’s a structural fit problem. Popunder shows the offer to a user who didn’t ask for it. That is exactly wrong for a $10K-ACV enterprise sale.

Where popunder actually wins

Impulse-friction offers. Deposit-bonus iGaming. Sweepstakes. Utility installs. Browser-extension and VPN offers. Anything where the user can convert in under 30 seconds with one form field, one click, and a price low enough that they don’t need to think.

The reason is mechanical: a popunder removes the publisher context. The user is no longer reading the article they came for. They are looking at your landing page in a fresh tab. If your offer is something they can evaluate in three seconds and click on immediately, popunder is the cheapest impression-to-conversion path in ad tech. If your offer needs five seconds of evaluation, popunder is wrong, and Google search is right.

In 2021 I ran an Italian iGaming campaign that booked €2.40 CPM popunder against an Adsy.tech-equivalent rate card. The same offer on Google Search ran €4.80 CPC with Quality Score 6/10. Cost per first deposit on popunder was €38. On Google Search it was €31 — but Google’s volume cap was 12 deposits per day. Popunder scaled to 80 deposits per day at the same cost-per-acquisition once we shut down three publishers whose traffic looked good in the panel and converted at 0% in CRM.

Where Google Ads wins

Considered-purchase offers. B2B. Enterprise software. Anything that needs the user to read three case studies before they fill out a form. Anything in a regulated vertical where Google’s Quality Score penalises bad landing pages enough that bad operators self-select out.

Google Ads also wins on geographic precision. You can target a single zip code in Manchester. Popunder networks let you target country and sometimes region. The granularity gap matters for local-service offers but doesn’t matter for global iGaming or LATAM dating.

Google wins on brand-safety theatre. If your CMO is going to read a leaked traffic source list out loud at a board meeting, you don’t want a publisher network in there. Google Display has the same long-tail problem, but the brand on top of it is one your CMO has heard of.

The hidden costs that make the comparison harder

The CPM you see in the popunder panel is not the CPM you get on conversion. The gap is typically 15–40%, depending on GEO and how aggressive the auction is. The reason is auction mechanics: most networks clear at second-price plus an increment, not at your bid. Some panels show your bid (the “ceiling”); some show the clearing price (the “actual”). Adsterra and PropellerAds historically showed different numbers. Adsy.tech shows the clearing price by default. When you compare networks, ask the AM which one they show. The answer separates the operators from the marketers.

Google Ads has the same hidden-cost problem, expressed differently. Your effective CPC depends on Quality Score, ad rank, and the auction. The number you bid is not the number you pay. The gap is usually 10–30%. Google’s panel does show the actual CPC paid — the bid is the ceiling, the auctioned price is the actual — but the multiplier effects of Quality Score across an account take three weeks of optimisation to settle.

Both formats also have data shrinkage between tracking and CRM. On popunder, I have personally watched campaigns convert at 12% in Voluum and 0% in the advertiser’s CRM, because a publisher in Q2 2022 was firing the conversion pixel from a server farm in Vietnam before the user actually saw the offer. It took six weeks to find. On Google Ads the equivalent failure is bot clicks — well documented in the 2019 IAB study at 26% non-human across programmatic. Server-side conversion validation is non-negotiable on either format.

A story about a $4.20 cheque

In January 2019 PropellerAds disabled the payout floor for one accounting cycle. The reason was a database migration. The consequence was that an account paid out a balance of $4.20 via physical cheque, mailed to my office address in Malta. The postage cost more than the cheque.

I am telling you this because it is the kind of operational economic that explains why network panels look the way they do. Payout floors are not a courtesy. They are operational economics. When they break, the math is absurd. When they work, they hide a lot of small fees, fractional cents, and rounding decisions that you wouldn’t tolerate from a SaaS pricing page. Networks publish rate cards that look clean and clear at the marketing level. The actuals are messier because the underlying market is messier.

The same is true for Google Ads. The CPC you bid clears against a Quality-Score-weighted auction, but the actual cost-per-acquisition you book reflects landing-page conversion rate, audience temperature, day-of-week effects, and a Quality Score that updates on a schedule Google doesn’t fully publish. Both formats look cleaner in the marketing copy than they do on a Tuesday afternoon when a campaign is misbehaving.

When you should run both

You should run both if you sell an impulse-friction product in a regulated vertical with a meaningful free trial. iGaming with a no-deposit bonus, VPN with a 7-day trial, dating with a free first message — in those cases Google Search captures intent (people Googling “best VPN 2026”) and popunder captures volume (cold tier-1 traffic at $0.40 CPM that converts at 0.4% but scales to whatever your daily cap is).

The two formats also play different roles in attribution. Google Search is high-intent, fast-converting, easy to attribute. Popunder is low-intent, slower-converting, harder to attribute. If you only run Google Search, your CAC looks artificially low because you’re paying for users who would have converted anyway. If you only run popunder, you’re scaling on impressions to people who don’t know who you are. The mix matters.

When you should run neither

If your offer needs five seconds of consideration before the click, popunder is wrong. If your CPC on Google Ads is over $20 because Quality Score is fighting you, Google Ads is wrong. The right format in that case is usually native ads, content marketing, or organic SEO — and I won’t pretend to be the best person to advise you on those, because I never ran native at scale.

What to do next

If you want to test popunder for an impulse-friction offer, the cheapest way to get a real number is to spend $50 on a network that publishes its rate-card minimum honestly. Adsy.tech does — $0.50 CPM, listed on the rate card. Adsterra and PropellerAds will negotiate down from a higher rate but you have to ask. If you want to test Google Ads for a considered-purchase offer, spend $200 and run it for three weeks before you decide anything, because the first three days of a Google campaign’s data is unreliable for the same reason the first three days of a popunder campaign’s data is unreliable: the system hasn’t trained on your offer yet.

If you’re choosing between formats for the same offer and your vertical fit isn’t obvious, my advice is to run both for two weeks with the same daily budget, hold attribution constant (one tracking platform, server-side conversion validation, CRM reconciliation), and compare cost-per-converted-user against retention at 30 days. Not at 7 days — sweepstakes audiences convert in week one and die in week four, and you will fool yourself if you measure too early.

See my honest 2026 popunder network ranking for which network to use, and the format glossary for the mechanics. Or if Meta Ads is the comparison you actually meant, here is the Meta version.

FAQ

Is popunder cheaper than Google Ads?

On CPM-to-CPC, popunder is cheaper for impulse-friction offers in tier-1 and most tier-3 GEOs. On cost-per-converted-user, the answer depends on offer fit. Utility installs and deposit-bonus iGaming are clearly cheaper on popunder. Enterprise SaaS over $1K ACV is clearly cheaper on Google Ads. Dating is roughly comparable.

Does popunder traffic convert?

For impulse-friction offers, yes — in the ranges I gave above (0.4% to 12% depending on vertical, offer LTV, GEO). For considered-purchase offers, no. Popunder removes the publisher context, which helps offers that need a quick decision and hurts offers that need slow trust-building.

Why does Google not allow iGaming in my market?

Google’s policy is country-specific and updates frequently. Italy, the UK, parts of LATAM, and some Nordic markets allow whitelisted operators. Most others do not. Popunder networks accept regulated iGaming traffic where Google does not, which is one of the structural reasons the format exists.

What about brand safety?

Popunder traffic comes from publisher networks where the bottom 30% of inventory is bot-adjacent or low-trust. Asking for “premium only” frequently means asking for the middle 60% of the network, because the top decile is reserved. Google Display has the same problem with less transparency. If brand safety is binding, run Google Search (intent-based, no display network) and skip both popunder and Google Display.

Sources. CPM ranges from Marco DeLuca’s PropellerAds account-book sampling 2018–2023 plus 2024–2026 buyer-side testing across Adsy.tech, Adsterra, and RichAds. Google Ads CPC ranges from Google Ads benchmark reports and Marco’s first-party Italian iGaming campaign data Q3 2023. IAB fraud rate from the 2019 IAB Tech Lab benchmark, iabtechlab.com. EDPB summary of EU popunder ePrivacy guidance, EDPB 2024 edpb.europa.eu.