160% ROI on $120 of Pops: Fact-Checking a Gaming Case Won After the Traffic Stopped

An affiliate buys 4 days of popunder traffic for $120, shows it only to Windows desktops with gaming graphics cards and sends it to a free-to-play football game that pays per active player. On the last day of traffic the campaign is down 57%. Then he does nothing for 2 weeks, and it finishes at 160% ROI.

That's the case Alukard_X posted on a forum on October 7 and PropellerAds turned into a blog post on October 19. This is our review for media buyers who run pops on install, gaming or any slow-converting offer: the author's thread, the network's version, the tracker's daily data, our own math and what neither version shows. Editorial fact-check: 3/5.

Short version: the money is real and reconciles across the tracker, the traffic source and both versions of the case. 130 of the 156 conversions really did arrive after the traffic was paused. But the test is $120, the offer's terms aren't public, the author links the network with a referral link, and the final GEO table in the network's post leaves out 17 of 37 countries, including 2 that were already in profit.

Who ran the UFL case and who published it

The author is Alukard_X, a forum regular with the "Super Affiliate" title, registered since October 3, 2020, with 1,075 messages. On October 7, 2026, he posted "Case Study UFL F2P football game" in the pops section of the forum, with the stack in the subtitle: "PropellerAds ➜ Keitaro ➜ Admachina • CPP • 37 GEOs • ROI +91.55%". He edited it on October 9. Logged-out readers see the setup, the money and the GEO split; the rest is for paying members.

Note the 3rd row: "Admachina (<- referral link)". If you sign up with the network from his post, he earns on it. He says so, which is more than most cases do.

12 days later PropellerAds published "Our Community Shares: How Alukard_X Hit 160% ROI With Zero Optimization [Case Study]". The post is signed by Olya Mikheeva, published October 19, updated October 20, and opens with "All materials and creatives were provided by the partner." It calls him "one of our longest-running partners" who has "been running traffic with PropellerAds for years", links his thread, the network's Telegram chat and an Admachina sign-up link with a referral code, and adds the final numbers to October 11.

Our take: this is a customer case packaged by the traffic source, but with an unusual paper trail. The author's own version came first, with his own numbers, and the network's version updates them rather than replacing them. Neither says whether the write-up was paid. Read it as a real campaign by a long-time PropellerAds customer, told in the network's marketing voice.

The setup: UFL, Admachina, 37 GEOs and 1 bid

From both versions combined:

• offer: UFL, a free-to-play football game, through Admachina;
• model: CPP, cost per player, paid for "login / active", "Varies" by GEO, "around €1.80 per player" on average per the network;
• traffic: PropellerAds Onclick, popunder and direct click, on SmartCPM with a max bid of $0.58 CPM, 1 bid for all GEOs;
• tracker: Keitaro;
• GEOs: 37 countries, Europe plus Brazil and the US;
• dates: traffic September 25 to 28, 2026, results counted to October 11.

UFL is a real, large game. Its official site says it's "available on PlayStation Store, Xbox Store, and PC"; the PC version launched in early access through its own launcher on November 7, 2025, and the Steam page still says "To be announced". The site's copyright line names XTEN Limited, which owns the STRIKERZ and UFL trademarks.

Admachina is less transparent. Its homepage calls it a "digital media agency for games and esports" offering "premium traffic for games" to advertisers and "instant payments" to webmasters. It names no legal entity, and the sign-up runs on Affise, a white-label affiliate platform. There's no public offer page with the UFL payout, the GEO list or the definition of "active".

Not shown: how the click reaches the game. Neither version says whether the pop opened the game's site directly or a pre-lander first, and there's no screenshot of either.

Targeting: the game's spec sheet as a filter

The 3 settings that the case credits for "zero optimization":

1. Devices. Desktop only, Windows 10 and 11, and graphics cards from 2 families: GeForce RTX and GeForce GTX 16. PropellerAds' help center lists "Graphics card targeting (available only for Onclick campaigns)", so this setup can't be copied to push or in-page formats.
2. Frequency. 1 impression per user per 336 hours, which is 14 days. The post says the network's default recommendation is 3 per 24 hours.
3. Exclusions. No VPN traffic, and nothing else: no browser, language, connection or zone limits.

Why graphics cards: as Alukard_X puts it in the network's post, "even with an RTX card, it's a 30 to 40 GB download, so only people who really want to play will go all the way."

Our math: the filter matches the game's minimum spec almost exactly. The Steam page lists "AMD Radeon RX 5600 XT or Nvidia GeForce GTX 1650" and 30 GB of storage. The GTX 1650 is the entry card of the GTX 16 family, so "RTX plus GTX 16" means every Nvidia gamer who meets the minimum. It also means no AMD cards, which also meet the minimum: an untested half of the audience, not a proven exclusion.

Our take: the 14-day cap is the smart part. It's the same length as the offer's 14-day cookie window, so every user gets 1 chance, and every conversion in the tracker belongs to a single pop. This isn't "zero optimization", it's optimization moved from the dashboard into the setup. The OS, the graphics card, the cap and the VPN exclusion are exactly the cuts most buyers make on day 3.

PropellerAds2.2

Onclick pops with graphics card targeting, used in this case

Visit site Read the review

4 days of traffic, 2 weeks of conversions

The most useful screenshot in the case is Keitaro's report by day. The traffic ran 4 days; the conversions kept coming for 13 more.

We grouped the same rows into 3 windows. Revenue is in euros, as the tracker shows it; ROI uses the euro-to-dollar rate implied by the case's own final figure (€277.76 = $311.54, or 1.1216).

Our math: 83% of the conversions came after the last paid impression. The campaign crossed break-even on October 1, 3 days after the traffic stopped. A buyer who judged it on September 28, as most of us would, saw $120 spent, about $51 back and a GEO list full of red, and would have killed a campaign that went on to return 2.6 times its cost.

There's no decay in the tail either. From September 29 to October 11 the tracker logged about 10 conversions a day, with 5 on the last day. The cookie window for a click on September 28 runs to October 12. So the final count likely missed a few late players, which makes the 160% conservative rather than inflated.

Why so slow: the network's post says a player must download the game, install it and launch it, which is "the first goal", then "actually play for a while", which is "the second one". The exact "active" rule isn't published. For a 30 GB download, a lag of days is believable.

The money: $120.04 in, $311.54 back

There are 3 sets of numbers, from 3 dates, and they line up.

Our math:

• October 7, forum: $229.94 minus $120.04 is $109.90, and $109.90 / $120.04 is 91.55%. The tracker's daily rows put that snapshot in the middle of October 7.
• Interim, PropellerAds: $137.383 / $120.037 is 114.45%, $120.037 / 270,854 impressions is a $0.443 CPM, and $120.037 / 141 is a $0.851 cost per conversion. All 3 match the screen.
• October 11, final: $311.54 minus $120.04 is $191.50, and $191.50 / $120.04 is 159.53%. The $311.54 is the tracker's €277.7645 at 1.1216 dollars per euro, close to the ECB's 1.1206 on October 9.

2 smaller things don't match exactly. The tracker reached 141 conversions on October 9, with about €253, or roughly $284, of revenue; PropellerAds shows $257.42 for the same 141. That's about 10% less, likely a payout passed by postback in a different currency or amount; the case doesn't explain it. And the post claims "90%+" unique hosts, while Keitaro shows 158,365 hosts on 186,709 clicks, or 85%.

Unit economics, from the final figures: 156 players at an average €1.78, about $2.00 each, which fits "around €1.80". Each cost $0.77. Revenue per 1,000 impressions was $1.15 against a $0.443 CPM. Break-even at that CPM is 0.22 players per 1,000 pops; the campaign made 0.58.

Note what the max bid did: $0.58 was a ceiling, not the price. On SmartCPM, per PropellerAds' help center, "automatic bidding is enabled by default" and "our system will set the optimal rate based on your targeting". The average came out 24% below the cap. To set your own ceiling, work back from expected revenue per 1,000 pops; our max bid calculator does the arithmetic.

Also worth knowing: only 186,709 of the 270,854 paid impressions reached the tracker, 69%. On pops that gap is normal, and the case's numbers already include it: spend is the source's, revenue is the tracker's.

GEOs: 2 snapshots and a table with gaps

The forum post has the GEO split on October 7. The network's post has a table "Final results by GEO", which its caption says comes from Alukard_X: spend from the PropellerAds export and revenue from his side.

Side by side, for the GEOs in both:

Our math: the final table lists 20 of the 37 GEOs, $95.40 of the $120.04 spend. The other 17 countries spent $24.64, brought $29.44 and made $4.80 together, and they're simply not shown. Among them are Albania and France, at $4.35 and $3.07 profit on October 7. With the traffic off, a GEO's profit can only grow, so both belong above several rows that made the table. So do Sweden, Austria and Switzerland, the losers the post's own text names.

That table also produces 2 claims that don't hold:

• the caption says "Top 5 GEOs: almost 90% of the total profit". The 5 highlighted rows add up to $144.02, which is 75% of $191.50;
• the text says "less than half of the GEOs were profitable". That's the October 7 figure, 16 of 37. By October 11 the table shows 17 GEOs in profit, and with Albania and France at least 19 of 37, just over half.

The post's GEO paragraph, with the US as the biggest loss "followed by Sweden, Austria, and Switzerland" and Albania in the top 10, also comes from the October 7 snapshot, not the final one.

Our take: none of this changes the result, it changes the lesson. The network frames the GEO split as "you pay a little to find out where the offer works, and the winners cover the rest." In the final numbers, the 3 losing GEOs on the table lost $3.74 combined; the wide launch was cheap because the losers barely lost. The real pattern is concentration: Spain and Italy took 15% of the spend and made 45% of the profit. Brazil took the most spend, 18%, at 48% ROI, and the US took 12% at -20%. 1 bid across 37 countries let cheap GEOs fill the budget and paid the same ceiling in Switzerland as in Albania. On a scaled run, a GEO split with separate bids is where the next 100 points of ROI would come from.

Can you repeat it in November 2026?

What you can copy:

• the setup, in about 10 minutes: Onclick, SmartCPM, Windows desktop, the 2 Nvidia families, a 1-per-14-days cap and no VPN;
• the test size: $120 over 4 days is a budget almost anyone can risk;
• the patience: count the result after the cookie window, not on the last traffic day.

What you can't:

• The timing. Alukard_X names it himself: "EA Sports FC 27 had just launched, and it costs $70 on its own." EA's store page confirms the PC release on September 25, 2026, the day his traffic started, at $69.99. A free football game had its best possible week. In November it's competing with a game people have already bought.
• The offer terms. Payouts by GEO, the "active" rule and whether conversions after the cookie window count aren't public. "Varies" is the only payout in the author's post.
• The scale. 1 run, 4 days, $120. Nobody has shown what happens at $50 a day, when the frequency cap and the graphics card filter start to limit volume.

Our take: before you test, ask Admachina, in writing, for the UFL payout per GEO, the exact "active" event, the cookie window and their rejection policy for late conversions. And track by click date, not just conversion date, so you can see which traffic day each player came from.

Fact-check: what the author showed and what they hid

We rate every case on 10 points: what an outsider needs to judge the result and repeat it. We scored the forum post and the network's post together. Like most case studies, this one is selectively true rather than fake.

Score: 3/5. The money is shown 3 times and reconciles every time, the targeting is complete, and the daily tracker data proves the case's main point: most of the result came after the traffic stopped. It doesn't go higher because the offer terms and the flow are hidden, the author earns from the network he recommends, the test is $120 and was run in the best week of the year for this game, and the network's final GEO table leaves out 17 countries and misstates its own top 5. Good data, small test: a 3, not a 4.