Three real e-commerce campaigns that bought EZmob popunder traffic spent $10,829 in ad spend and returned $18,487 in tracked revenue — a blended 71% return, with the best single campaign (a DoorDash app-install offer in Canada) clearing +150% ROI at a 1.5% click-to-install conversion rate. All three were published as EZmob case studies, submitted by the media buyers who ran them, with the spend, revenue, bids and targeting left intact. This page assembles the three into one playbook you can copy in 2026.
- Three published EZmob e-commerce campaigns spent $10,829 and returned $18,487 — a 71% blended ROI.
- Every campaign started at the $0.20 CPM minimum bid, then scaled the placements that converted.
- DoorDash CPI in Canada converted 1.5% of clicks into installs for a +150% ROI.
- The shared playbook: launch broad at the floor bid, prune placements fast, segment supply with your account manager.
The three campaigns at a glance
| Campaign | GEO | Format | Spend | Revenue | ROI |
|---|---|---|---|---|---|
| German shop rotation | Germany | Popunder | $3,500 | ~$7,000 | +85% |
| Lazada app installs | Indonesia | Popunder | $1,645 | $2,961 | +85% |
| DoorDash app installs | Canada | Popunder | $5,684 | $8,526+ | +150% |
| Total | — | — | $10,829 | $18,487+ | +71% blended |
One methodology note before the detail: the figures come from the three case studies as published on EZmob's blog, submitted by the buyers themselves. The campaigns ran in 2022, so read them as proof the playbook converts on this inventory — not as current 2026 rate cards for your GEO.
Campaign 1: German shop rotation on popunders — $3,500 in, about $7,000 back
The first campaign promoted a German affiliate program that rotated about ten German-native shops — CBD, liqueur, home improvement, perfume and electronics — through a single affiliate link. The payout was a hybrid: a low up-front payment plus a revenue share on sales generated by newly registered users. Because the up-front payout was small, the buyer started at the lowest popunder bid available, $0.20 CPM, and treated the campaign as a list-building exercise: whitelists for placements that converted, blacklists for publishers that did not.
How the campaign was set up
- Format and GEO: popunder traffic, Germany only.
- Bid: started at the $0.20 CPM floor, raised only where a placement showed potential.
- Funnel: the shops' own landing pages were the creatives — no pre-landers, no ad copy to test.
- Optimization: after early traction, the buyer worked with EZmob's account managers to split the supply into higher-quality and lower-quality feed buckets, with a different bid on each bucket.
The supply segmentation was the turning point. Once feeds were bucketed by quality, the campaign opened up to better inventory, the conversion rate climbed, and the buyer duplicated the winning campaigns onto the individual shop pages that converted hardest — CBD and perfume among them — squeezing the rest of the offer's life for maximum profit. The full breakdown is in the German e-commerce popunder case study.
The lesson
Low-payout hybrid offers are a margin game, and popunders fit them because the floor bid stays cheap while inventory stays deep. The buyer's own rule of thumb: popunder bids under $0.30 CPM usually leave enough room for error to build whitelists before the revenue share element has to carry the campaign.
Campaign 2: Lazada app installs in Indonesia — $1,645 spent, $2,961 back
The second campaign was a cost-per-install push on Lazada, one of the largest e-commerce platforms in Asia, run through its affiliate program in Indonesia. Spend was $1,645 and tracked revenue $2,961 — $1,316 profit, the +85% ROI the published case study headlines. The effective cost per install averaged around $0.20 for the life of the campaign.
How the campaign was set up
- Format and GEO: popunder inventory, Indonesia in focus (several countries enabled, Indonesia the one measured here).
- Targeting: Android mobile only at launch — the lowest-cost, highest-availability slice — with carriers included in the mix because the app's download weight was small enough not to choke carrier connections.
- Bid: the same $0.20 CPM floor, which opened enough placements to discover performers quickly.
- Creative: Lazada's affiliate team supplied localized landing pages, banners and smart links, so the brand's own recognition did the converting on run-of-network traffic.
Conversions arrived from day one, and that early positive signal is what let the buyer keep the discipline: placements that did not perform got repriced within the first month, and the worst were blacklisted across every campaign on the account, not just this one. Details are in the Lazada CPI case study.
The lesson
A recognizable app brand converts on cold pop traffic when the install is cheap and the funnel is one click long. The job at launch is not creative genius — it is discovering which placements deserve the budget, which is a bid-and-prune problem, not a design problem.
Campaign 3: DoorDash app installs in Canada — $5,684 spent, $8,526+ back
The third and largest campaign bought installs for DoorDash's Android app as the brand pushed into Canada. Spend $5,684, revenue $8,526 and climbing — a +150% ROI — with the click-to-install conversion rate stable at about 1.5% across the entire run.
How the campaign was set up
- Targeting: Canada only, Android only, and — the decisive filter — AT&T carrier traffic only. The campaign met its daily caps every day without ever opening broader targeting.
- Bid: $0.20 CPM floor again, with a couple of days of clean data collected before optimization began.
- Funnel: none. Traffic went straight to the app store page. The conversion rate was already at benchmark, so the buyer deliberately skipped pre-landers and A/B testing and let the offer run.
- Optimization: once each publisher feed had spent roughly $200-300, the buyer weeded out sub-placements (SUBIDs) with one simple rule — about $10 of spend should produce one conversion. A SUBID that passed that amount without converting was either duplicated into a lower-bid campaign or removed entirely.
The lesson
Brand-heavy CPI offers can carry a straight-to-store funnel on pop traffic, and it was carrier targeting plus the $10-per-conversion rule that protected the margin. Two findings are worth stealing outright: older Android OS versions out-converted the newest one, and browser-level conversion rates differed enough that per-browser bids would have added profit. The full data — conversion rates by browser, OS version and device brand — is in the DoorDash Canada case study.
What all three e-commerce campaigns share
Strip away the GEOs and offers, and the same five moves produced all three results:
- Enter at the floor bid. All three campaigns launched at $0.20 CPM. Cheap impressions at launch are not a sign of low quality — they are how you afford discovery across hundreds of publisher feeds.
- Launch broad, prune fast. None of the buyers started with tight targeting. They started run-of-network on Android mobile, collected data, then cut. The DoorDash campaign narrowed to one carrier only after the data said so.
- Use one arithmetic pruning rule. The DoorDash case made it explicit: roughly $10 of spend per expected conversion. Any placement past that threshold without a conversion gets repriced or removed.
- Work the supply side with your account manager. The German campaign's profit inflection came when EZmob's team helped bucket feeds into quality tiers with separate bids. Account managers exist for this — the buyers who used them scaled faster.
- Keep the funnel short. No pre-landers in any of the three campaigns. Two went straight to the destination page and still converted at or above benchmark.
How much to trust these numbers
Every figure on this page comes from case studies published on EZmob's blog and submitted by the media buyers who ran the campaigns — the same numbers appear in each study's summary panel (GEO, format, cost, revenue). The limits are real: the campaigns ran in 2022, the offers and payout structures were specific, and none of the buyers published their exact CPM curves mid-campaign. Treat the results as directional proof that popunder inventory converts e-commerce and CPI offers profitably when the entry bid is low and pruning is aggressive — and treat your own first campaign as the benchmark that matters.
Running the same playbook in 2026
The setup these buyers used is still the fastest route into this traffic in 2026:
- Register an advertiser account — no sales call needed on the self-serve platform.
- Fund a test from a $100 budget, the platform's test budget scale.
- Launch a popunder campaign at the floor bid with broad Android mobile targeting.
- After a couple of days of data, apply the $10-per-conversion rule to every placement and SUBID.
- Bring your account manager in to segment supply by quality, then duplicate the winners onto your best-performing landing pages.
That is the whole method — the same one that returned $18,487 on $10,829 across these three campaigns. You can see the platform, formats and targeting options on EZmob's self-serve advertising platform before you spend anything.
FAQ
Do popunder ads actually convert e-commerce offers?
Yes — three published EZmob e-commerce campaigns spent $10,829 and returned $18,487 in tracked revenue, a 71% blended ROI. The largest returned $8,526 from $5,684 of spend at a +150% ROI.
What bid should you start a popunder campaign at?
$0.20 CPM, the floor all three case-study campaigns launched at. Starting low opens hundreds of placements for discovery, and you raise bids only on placements that convert.
Can popunder traffic deliver app installs for CPI offers?
Yes. The DoorDash campaign in Canada converted about 1.5% of clicks into installs, and the Lazada campaign in Indonesia averaged roughly $0.20 per install for the life of the campaign.
What targeting works best for e-commerce on pop traffic?
Launch broad run-of-network on Android mobile, then prune. Carrier-level targeting carried the DoorDash campaign in Canada, and older Android OS versions out-converted the newest one in the same campaign.
How much budget do you need to test popunder traffic?
EZmob’s self-serve platform runs tests from a $100 budget. The Lazada campaign spent $1,645 total and converted from day one, so a small funded test gets a real signal fast.
When should you blacklist a popunder placement or SUBID?
Use the DoorDash rule: about $10 of spend should produce one conversion. A SUBID that passes that without converting gets duplicated into a lower-bid campaign or removed entirely.
One last thing
The most counterintuitive number in the DoorDash study: the newest Android OS version was not the best converter — older versions delivered an unexpected share of both conversions and conversion rate. Before you default to targeting only the latest OS, let your own data make that call.