Build Your First AI Project This Weekend
Stop consuming tutorials. Start creating. Get the free step-by-step guide.
2026 Update:
PopAds is still a popunder-only network. Tomksoft S.A. has run it since 2010. Advertiser entry is still cheap: $10 deposit, $2.50 a day to run a campaign (AffMaven, PopAds).
It will not make you rich. It is still one of the cheapest places to practice bidding, targeting, and blacklists with real money.
It is not a customer channel for a dentist or an HVAC company. Those businesses should be on Google Ads and local SEO.
Why start here at all
Google Search CPCs for dentists sit around $8.00 and home improvement around $8.33 in WordStream’s 2026 search benchmarks (WordStream). That is the right spend if someone is searching for you.
PopAds is the opposite. Low intent, low cost, fast feedback. You learn:
- How a second-price auction actually bills you
- Why “cheap CPC” is often leftover junk
- How to cut site IDs instead of “optimizing vibes”
- How landing page language has to match the visitor
Those skills transfer to PropellerAds, native, and even Google. The traffic quality does not.
What PopAds is
Self-serve popunder RTB. A publisher runs their code. A visitor clicks something. A window or tab opens with your URL. You bid per thousand views (CPV-style popunder billing), not for a search keyword.
Why affiliates still use it:
- Tiny deposit
- City, language, OS, ISP, resolution, category targeting
- No account manager lecturing you
- You lose money when you are wrong (that is the lesson)
The catch:
- Blind: you get website IDs, not domains
- Quality follows bid. Low bid = leftovers and bots. PopAds says this in their own docs. Believe your conversion data, not their FAQ.
- You are bidding against a lot of other affiliates in popular geos
For the wider format, see pop traffic.
How the click actually happens
Mobile
True popunder. The site they wanted stays on top. They might see you hours later. “Only 5 left!” is a joke when they find the tab on Thursday.
Write for delayed attention. Evergreen hook. No fake countdown.
Desktop
Two flavors: a window, or a tab-up in the foreground. I have seen the same offer flip from green to red just by switching format. Test it. Do not assume.
Desktop is unfashionable. Older sweepstakes traffic still lives there.
Where the inventory comes from
File sharing, image hosts, browser games, auto blogs (scraped / spun content that rides social traffic).
You will not see PopAds on a hospital website. Good. You are not building a brand here.
Practical split I still use:
Start with general/news and general/society style categories (on PopAds these have been 100 and 127). Exclude file sharing and image hosting (250 / 251) until you have converting campaigns.
Auto blogs leak leftover social traffic without Meta’s ad policies. That is where I have found most winners. Torrent traffic is volume, not buyers.
Research before you deposit
PopAds publishes a traffic inventory report. Use it.
Geo prompt
I am researching PopAds in [COUNTRY]. Here is yesterday's inventory dump:
[paste]
1. Languages to target
2. Device / OS mix
3. ISPs and what they imply
4. Red flags
5. Starting targeting for a simple SOI sweepstakes
Language mismatch (this still works)
Belgium is the example I have used for years. A lot of affiliates default to Dutch. A lot of the traffic is French. Match browser language to the page. Always.
I am targeting [COUNTRY] on PopAds.
What languages should the landing page be in, in what order?
Which language do internet users actually use in the browser?
Small edge. Compounded across a week of spend, it is the difference between “this geo sucks” and “I was on the wrong language.”
Competitive research
Adplexity for live pop campaigns. Steal layout and language. Do not steal fake virus pages.
Bidding: the low-bid trap
PopAds is second-price. Highest bid wins, pays something closer to the next bid.
If you bid $0.005 because you want “cheap traffic,” you sit behind everyone who already blacklisted the garbage. You get:
- Their leftovers
- Users who already saw 12 pops
I start high, get clean-ish data, then cut down.
Even with a high max bid, you often pay far less than the cap. Check their estimator. Pass the [BID] token so your tracker records actual cost, not max bid:
https://tracker.example/click.php?c=1&key=YOURKEY&cpc=[BID]
Lesson that transfers to every auction: optimize for data quality first, cost second. This is true on Google too. The difference is Google traffic can pay a dentist’s rent.
Targeting that is worth using
Categories. Auto blogs first. File sharing later, if ever.
Language. Match the page. Bonus: a lot of bots do not bother changing language headers.
OS. Android 3.x is not a customer. Target current versions unless you have a reason. Ask your affiliate manager which OS versions convert for the offer.
Browser. Edge on desktop has converted sweepstakes for me. I start by excluding Firefox and Opera Mini on pop SOI. Test, do not treat that as gospel.
ISP / carrier. In the US, carrier is a rough income proxy. Match the prize to the carrier. Walmart card vs “luxury” prize is not the same audience.
Time. Impulse offers often work better evening and overnight local time. Ask the AM if they have converting hours. Then verify in your tracker.
Websites. Off at launch. After data, geo-specific blacklists. A site that is trash in Belgium might work in France. Share lists with people you trust.
Blacklist process that does not lie to you
- Launch with website targeting off. You need a sample.
- After 48 hours, export site ID, spend, conversions, revenue.
- Instant blacklist: spent more than ~3x payout, zero conversions, and enough clicks that “maybe it converts on click 80” is cope.
- Watch list: spend with 1 conversion. Do not whitelist one lucky click.
- Whitelist campaign: only after a site has enough conversions that you would bet next week’s budget on it.
- Keep lists per geo. Do not paste a US blacklist into Brazil.
PopAds quality scores ([QUALITY]) are a hint. Your tracker is the vote. I have seen “high quality” IDs dump bots and “low quality” IDs convert SOI. Filter on money.
Tokens to put in the offer URL
| Token | Why you care |
|---|---|
[WEBSITEID] | Blacklist / whitelist |
[QUALITY] | Compare against actual CPA, do not worship it |
[ADBLOCK] | Split if your page fights blockers |
[BROWSERNAME] / [OSNAME] | Cuts, not vanity |
[ISPNAME] | Carrier / income proxy |
[BID] | Actual cost, not max bid |
[SCREENRESOLUTION] | Bot clusters often repeat the same resolution |
If you only pass website ID, you can still work. If you pass none of them, you are guessing.
AI on the optimization loop
Export from the tracker. Do not stare at Excel for three hours.
PopAds last 7 days:
[site ID, impressions, clicks, conversions, cost, revenue]
Target CPA: $[X]
1. Blacklist (high spend, no conversions)
2. Watch list (some conversions, not enough data)
3. Whitelist candidates
4. Patterns in converters
5. Bid changes
Then combinations:
Break down by browser, OS, resolution, hour, weekday.
Which combos beat average CPA?
What should I cut?
I have found pockets like “current Android + Chrome + evening” that beat the campaign average. You will not see that in the network dashboard’s default view.
Kill / keep framework when you are in the red but improving:
5 days, $500 spend, $420 revenue, day 1 -40%, day 5 -8%.
Keep optimizing, pause, or kill?
What extra data would change that?
AI does not decide. It forces you to write the decision down.
Landing pages
Pop needs iteration, not a designer on a two-week cycle.
SOI page prompt:
Mobile landing page, [COUNTRY], [LANGUAGE].
Offer: legitimate iPhone / gift-card sweepstakes.
No fake urgency, no "you already won," clear rules link, email capture, trust line.
Output HTML/CSS I can host.
App pre-lander:
Pre-lander for [APP] in [COUNTRY].
Benefit, device compatibility, store CTA.
No virus or battery scare.
Mobile HTML + simple device-detection JS.
Alert box
On mobile the popunder sits behind the main window. An alert() yanks it forward. Aggressive. Helps some offers, hurts others. Split 50/50 in the tracker. Do not ship it as default.
alert("Your results are ready.");
Keep the copy honest. “Congratulations you won” is how accounts die.
Post-click rules
Do not send every click to one offer. Route by OS version, geo, hour, carrier in the tracker. Same skill you will use on every other source.
What works vs what is dead
Works (simple, low friction):
- Real sweepstakes (SOI)
- Legit app installs
- Short lead forms / quizzes
Dead or stupid in 2026:
- Fake AV / virus warnings
- Fake countdown “only 3 left”
- Hard sells that need a purchase on page one
- Local service lead gen
Pop visitors were not shopping for a crown or a water heater.
Reality: placements die, competitors outbid you, offers get pulled. You test, cut, replace. AI shortens the cycle. It does not remove it.
First week
Day 1: $100-200 in the account. Tracker live (Voluum or RedTrack). Geo research from the inventory report.
Day 2-3: Offers from your AM. Spy + AI. Two or three pages.
Day 4-7: $20-50/day. Daily export into Claude. Blacklist fast.
Week 2: Geo blacklists, micro-segments, kill or scale.
Do not “set and forget.” That is how you donate.
What I would not do on PopAds in 2026
- Bid the floor in the US and expect “cheap Google-quality traffic”
- Run a dentist, HVAC, or lawyer page
- Use one landing page in English for all of Europe
- Trust an affiliate manager who will not name converting site types
- Scale a campaign that is only profitable because of one unrepeatable site ID (clone it to a whitelist, then test if volume holds)
PopAds support is not going to build your funnel. The inventory report plus your tracker is the product.
Where to go next
Local business that clicked this from Google: you want Google Ads and local SEO.
Have questions? Contact me.