PWA services for gambling traffic: are they dying in 2026?

If you buy gambling or betting traffic, "PWA" does not mean what it means to a web developer. It means a service that takes an offer link and produces a page styled like a Google Play listing, with an icon, fake reviews, a download count and an Install button that adds a shortcut to the home screen and asks for push permission. You rent it per install or per month, point Facebook or TikTok traffic at it, and swap the domain when it gets banned.

There are more of these services every quarter. There is also less life in them than the listicles suggest. This is a count of both, done on 7 September 2026, followed by what two security companies found at the bottom of the same funnel in the same summer.

Why the category exists

The stores will not carry the product. Google Play has not approved gambling from unlicensed operators for years, and from 30 September 2026 Android's developer verification starts in Brazil, Indonesia, Singapore and Thailand, so even an APK installed outside Play needs a verified human behind it. A PWA is not an APK. It installs from the browser, needs no store and no verification, and can be rebuilt on a new domain in minutes. That is the whole pitch.

Meta's gambling policy, last changed on 18 May 2026, allows gambling ads only from accounts that obtained authorization and proved a licence for every territory they target, and it explicitly covers landing pages that promote gambling "even if there is no opportunity to gamble or game directly on that page, such as aggregator or affiliate sites". A PWA store page is exactly that kind of landing page. Which is why the industry guides pair it with cloaking. One guide published on 9 January 2026 tells readers to keep five to ten deployed copies on different domains, swap the URL in the ad account when a ban hits, and show reviewers a white page while real users get the casino. Read that as a description of a burn rate, not a strategy.

The mushrooms, dated

We assembled a list of PWA constructors from three directories for affiliates and from the services' own "top 5" posts, then looked up when each domain was registered.

Service	Domain registered	Public Telegram channel	Last post	Posts, last 30 days
PWA.Group	April 2021	1.9K subscribers	4 Sep 2026	3
PWA.Market	October 2021	1.36K	28 Jul 2026	0
ZM Apps	October 2023	3	2 Aug 2025	0
BlackApp	October 2023	680	4 Aug 2026	0
PWA.bot	November 2023	support bot only	n/a	n/a
PWA.Partners	May 2024	894	28 Aug 2026	3
Betterlinks.pro	April 2025	3	9 May 2026	0
Adset.pro	October 2025	personal contact only	n/a	n/a
EpicPWA	November 2025	no public preview	n/a	n/a
Devils PWA	Telegram only	669	2 Oct 2025	0
PWA OneTraff	Telegram only	163	21 Apr 2026	0
Meow PWA	Telegram only	33	23 Jun 2026	0

Two services that appear in current "best PWA builder" lists did not answer at all: the site behind Kelnik did not resolve, and CPA.STORE's channel last posted in April 2024. Comsign PWA has a live site but no public channel to check.

Read the dates downward. Of the 2021 pair, one still posts weekly and the other went quiet at the end of July. Of the 2023 cohort, one channel has three subscribers and last posted a year ago, one posted once in early August, and one has nothing but a support bot. The 2024 entrant is the only younger service still posting. The 2025 cohort arrived with the vocabulary of the AI boom: Adset.pro, registered in October 2025, sells "five tools in one account" with a "15-minute setup" and calls itself an AI-first platform; EpicPWA, registered in November 2025, sells "AI features" and a free tier. On the directory that lists app-rental services for media buyers, 13 of the 30 entries carry logo files uploaded in a single month, December 2025.

Meanwhile, the review pages. Search "EpicPWA review" and you get five near-identical articles on five affiliate blogs, each with a disclosure that links earn a commission. The one we read in full was published on 7 May 2026 and quoted 2,488 active clients, 1,524 apps and 12,484 installs. On 7 September 2026 the service's own homepage showed 2,488 affiliates, 8,124 apps and 129,484 installs. Apps up fivefold, installs up tenfold, clients unchanged to the digit. Either nobody new signed up in four months while installs grew ten times, or the counter is decoration. Either way, the "review" is a product page with a different domain.

The channels that still talk

We ran the same check on 16 services that rent native iOS and Android apps rather than PWAs, from the same directory. Eight of them had posted in the previous 30 days, some every few days. Of the nine PWA services with a public channel, two had. The people renting native apps are busy. The people renting PWAs, with two exceptions, have gone quiet.

That is the honest shape of the market: a technique everyone talks about, sold by a handful of survivors from 2021, surrounded by shells that were opened in 2025 and never filled. The demand did not vanish. It moved, and the next section is where.

What the security industry found at the end of the funnel

On 7 September 2026 Netcraft published its analysis of what it calls branded gambling campaigns. Paid ads on Facebook, Instagram, Threads and TikTok borrow the logos of Monzo, Revolut, Barclays, Tesco, Amazon and Netflix. The click lands on a page that "imitates a Google Play Store or Apple App Store listing", with the brand's logo as the app icon, a fabricated developer name, fake reviews and fake download counts. Tapping Install, in Netcraft's words, means "no app is downloaded from any official store. Instead, a Progressive Web App (PWA) is installed". The economics are stated plainly: typical cost-per-acquisition payouts of $50 to $350 per depositing player.

NordVPN's threat intelligence team got there first. Its findings, reported on 31 July 2026 and in more detail on 7 August, describe a campaign it labelled "pwa_betterlinks": more than 400 impersonated brands, about a thousand affiliate accounts across more than 250 teams, 3,100 fake pages served to ad reviewers and 7,200 user redirection records. At the top of it sits a platform that, in NordVPN's description, markets itself openly as a "PWA constructor for affiliates" with built-in cloaking and Facebook optimisation. That platform's domain is in our table above, registered in April 2025. Its public channel has three subscribers. Allure Security, writing on 30 July 2026, described the same kit: it detects Android or iOS and shows the matching fake store page, and "could be reconfigured through simple settings changes to present different casino names, fake reviews, and developer identities".

Put the two halves together. The public PWA services are thinning out. The volume is in a private kit rented by a thousand accounts, and three security companies have now written it up with domain names. Google Ads files this under unacceptable business practices, where making it seem you are affiliated with another brand is an egregious violation with immediate suspension and no warning. Meta's policy language for the same page is "scam ad". A traffic method that has become a security-industry case study is not a method with a future.

So, are PWA services dying?

As a business you can find and pay for, yes. The category peaked as a Telegram-first cottage industry, is consolidating to the two names that predate it, and the fresh entrants are AI-flavoured shells and affiliate review farms. The technique itself is not dying. It got worse: it became a brand-impersonation kit, and it took the "PWA" label with it. Anyone typing "PWA for gambling" into a search engine in 2026 is now reading, in order, an affiliate review, a cloaking guide and a Netcraft report.

There is a sharper reason the service model is over. A PWA store page is a manifest file, one HTML template and a push script. In 2026 that is a prompt, not a product: describe the page to Claude Code, Codex or ChatGPT and it writes the manifest, the install button and the push subscription in one sitting, with no account at any service. Lovable alone reported about a million new projects a week in June, with 80% of builders calling themselves non-technical. A media-buying team that wants a fake store page does not need to rent one at $25 a month. It generates one, which is exactly what makes every copy indistinguishable to Meta's reviewers and every domain disposable. Nobody needs the service, because the service was never the hard part. The hard part was surviving review, and nothing in a PWA helps with that.

What to do instead

If you run a licensed product, the honest version of "install" still exists: a real PWA of a site people already use, installed by people who came looking. We run one ourselves and it is ten seconds to install, with nothing to cloak because there is nothing to hide. Reach that survives review comes from an authorised ad account with a licence per market, from Microsoft Advertising, where gambling is allowed with documentation, and from communities where a real person wrote a real opinion. That last one is what AntiMedia's tasks for authors produce, and it is the opposite of a page with fake reviews.

If you run an unlicensed one, the numbers above are your forecast. Five to ten domains per campaign, a kit that a thousand other accounts also rent, and a research team publishing your infrastructure by name. The ad bans article explains where regulated brands went when this door started closing; the Android verification piece covers the door that closes on 30 September.

Method note

Services were collected from the PWA and app-rental sections of two affiliate directories and from a PWA vendor's own "top 5" post dated 20 March 2026. Domain registration dates come from RDAP lookups on 7 September 2026. Telegram figures come from the public web preview of each channel on the same day: subscriber count, date of the newest post, and the number of posts dated after 8 August 2026. Channels without a public preview, personal contacts and support bots are marked as such rather than counted as silent. Site availability was checked with a desktop browser user agent. The raw table is available on request.