An Indian father was charged ₹1,100: CountAI brought the domestic "photo-count pipes" idea to India and is paying back through YouTube ads and weekly subscriptions

CountAI is a photo-counting tool on Google Play: photograph a pile of pipes, parts or crates and AI tells you how many there are. Launched on Jul 21, it had 242k cumulative installs by Nov 6, charting only in the India Tools Top Grossing category, at #86. It has no new technology. Overseas similar products have existed for nearly a decade, and at home the category is long crowded. It worked because of a going-global performance marketing company's established playbook: pick India, buy YouTube ads, and hard-gate the paywall into weekly subscriptions. Below we cover how it got started, how the money is collected, and whether you should build one yourself.

A one-star review that starts with a YouTube ad

On Sep 2026 16, an Indian user left a one-star review on CountAI's store page: "My underage son saw this app in a YouTube ad. After downloading he paid a small upfront fee and didn't know it would turn on auto-renewal. A UPI auto-debit mandate of up to ₹1,100 was set up this way, and ₹1,100 was taken from my bank account. I've applied for a refund. Parents should be careful with this app's subscription and auto-debit terms."

A month earlier, on Aug 19, another one-star review was shorter: "Scam app that charges money, wants to set users up for a ₹1,100 weekly auto-debit. Don't download."

These two reviews pretty much sum up CountAI: users come from YouTube ads, enter through a small upfront payment trial, and the trial automatically converts to a weekly subscription through India's UPI auto-debit.

What it actually is

The store page says it's a "multi-scenario AI scanning counter that can change counts, add notes, and save and share results." The description claims several "in-house" recognition models for ordinary goods, loose parts, stacked materials and small accessories, aimed at warehouse inventory, construction material estimates and office tallies. In-app purchases in the India store run from ₹850 to ₹5,600, and in the US store from $7.49 to $49.99 (Google Play, 11-06).

The developer is FuMaoMao. On the store, the entity is FU MAO MAO PTE. LTD. in Singapore, while the iOS version is registered to Shaoxing Changen Network Information Technology Co., Ltd. Whether these are the same people can only be inferred from identically named apps. There is no direct evidence. This developer only published its first app at the end of April 2026. Now, 10 apps have accumulated about 10M installs: BabyGenie 4.67M, HairGenie 1.8M, DancePop 1.25M, and a screen recorder at 1.2M. CountAI is another category on this same production line.

How the 240k installs happened

The install curve estimated by third parties has a very typical "switch" shape:

Products that grow through search or word of mouth don't look like this. They hit full volume two weeks after launch, shut off abruptly a month later, and then turn on and off in bursts. The most likely explanation is that it spent four weeks of budget to check payback, then bought in an on-and-off pattern depending on how payback looked (inferred). One more detail: the version stopped at Sep 4. Once the surge period ended, the product was never updated again, and only the ad buying has been adjusted since.

The channel is essentially certainly Google App Ads. The app appears in YouTube ads on Play, which is the only route for that. The reviews are full of rupee amounts, Indian-style English and Hindi ("ok bhaiya thanks", "use nhi kiya krke duga review"). Its only charting was in India's Tools Top Grossing category, at #104 at the end of October and #82 in Nov 5. Nobody is discussing it on TikTok, X or YouTube, and it doesn't show up in the public Meta ad library, though that doesn't mean it hasn't run Meta ads.

4.67 stars from 7,822 ratings look very good, but only about 393 of those ratings have text, and 17% of the written reviews are one-star. The ratings without text most likely came from the in-app rating prompt (inferred). Real satisfaction is much lower than the 4.7-star average suggests.

How it makes money

There are no free uses. You pay as soon as you open it, and twenty to thirty reviews complain about this: "Can't do anything without a subscription," "Even a three-day trial costs money," "Its only job is counting things, and it charges you to count."

The monetization pattern is spelled out more clearly in its sister app. BabyGenie's reviews mention "₹10 trial for 3 days," which auto-converts to a weekly subscription. The figures visible in CountAI's reviews are "₹1,100 per week" and "₹600 down the drain." India's UPI auto-debit mandate, set up once, can charge every week, so an impulse payment of ₹10 becomes a string of weekly fees. Most of the cost of refunds and complaints is absorbed by users themselves.

The math works because installs are cheap in India. A single install from tool-app ads in India typically costs a few cents to about 30 cents USD (estimated, unverified). If a few thousandths of installs convert into a weekly subscription worth ₹850 or more, it pays back. Revenue for the Indian Tools Top Grossing chart's #80 to #100 positions is roughly several hundred dollars a day (estimated, unverified). That isn't big money, but for an app built on a reused template that is no longer updated, it's enough. The cost is reputation: "scam" and "fraud" are everywhere among one-star reviews. This playbook never relied on reputation in the first place.

The demand itself is real. On Aug 3, a four-star review said: "I think it suits students and wholesale and retail shop owners." Counting pipes on a construction site, goods in a warehouse, or stock in a small shop: getting the count right once saves half an hour, so some people will pay even for only one week.

Who it's competing with

The category isn't new overseas, and it isn't short of players:

The last row is the deadliest. On Aug 19, one one-star review said in a single line: "Paid app. Gemini does this for free. A very silly app."

At home it's even more crowded: point-counting cameras, photo-counting kings, color-bean counting cameras, pipe counters, smart counters. The app stores have dedicated collections, and people have long paid to count pipes on construction sites and do warehouse inventory. Some even advertise "free and unlimited." CountAI's description (pipes, parts, seed-specific models, changing counts, adding notes, exporting) is almost an English version of these products.

Where its real difference lies

There's almost none in the product. Its recognition isn't better than Google Lens, its features are a copy of domestic peers, and it's less complete for B2B than CountThings. On Aug 11, a one-star review that 16 people marked "helpful" said: "Don't use it for serious work. It even miscounts the basics. It's okay for teaching kids to count."

The difference is in the playbook. Overseas peers focus on Europe and the US. CountAI went to India, where installs are cheap, UPI supports weekly debits, and there are many construction workers and small vendors. Its paywall is harder than its peers', and its trial is "smaller." Behind it is a performance marketing company that already proved the "India ad buys plus small trial" payback model with BabyGenie. CountAI is just one more category hung on that same machine.

Few people know about it. There are no public revenue figures, no media coverage, and zero discussion on social platforms. But the category itself is no secret: within three months, followers like SaverHub had already appeared.

If it were you

Verdict: Building another English counting app like CountAI isn't worth it. What's worth learning is the product-selection method behind it: take small tools that people already pay for at home, localize them, and buy ads for them in India and Southeast Asia.

Who it suits: Small going-global teams that already know how to run Google App Ads and have a subscription template and payback model. What you need isn't a good product, but a acquisition workflow that tells you within four weeks whether it pays back.

Who it doesn't suit: Indie developers who don't know how to buy ads and are counting on store search or content to bring users. This category is squeezed by the free Google Lens: without spending on ads you won't get a single download. It also doesn't suit people who care about reputation and long-term retention.

How to do it: Don't crowd into "counting things". Go through domestic app stores for tools in the same kinds of scenarios that haven't gone overseas yet: photo distance measurement, volume calculation, photo-based material identification, counting timber, tile quantity estimation, counting pills. Make an English version, connect Google App Ads, and buy traffic in India and Southeast Asia. You can borrow the paywall, but the trial terms must be clear, and the auto-debit amount must be disclosed before payment. Otherwise one-star reviews and refunds will grow faster than revenue, and there's a risk that Google Play will treat it under its subscription policies.

How to acquire users: Google App Ads, mainly YouTube placements. Run four weeks in India first, and watch the cost per install and first-week subscription conversion. At the same time, build store-page keywords around long-tail terms like "count objects" and "pipe counter", and add some organic traffic.

When to move early: The window for any single category is short. CountAI had copycats within three months of gaining traction. The category you pick should go from launch to the first round of ad results within two or three months.

Signs you've misjudged: After four weeks of the first round of spending, the cost per install won't come down below a few cents to about 30 cents USD, or the share of first trials converting to weekly subscriptions is too low to pay back. Or Google Lens and Gemini have already handled your scenario well enough that user reviews start comparing your app to them.