← Back to blog

Case Study: An Android App on Telegram Mini Apps — 384 Conversions at $1.25 CPA in Four Weeks

An Android video app ran two video campaigns on Telegram Mini Apps in the US and Germany. October 2026: $481 spent, 150,387 impressions, 21,700 clicks and 384 tracked conversions. Conversion rate held at 1.7–2.0% every week; what moved CPA was click-through rate.

Case Study: An Android App on Telegram Mini Apps — 384 Conversions at $1.25 CPA in Four Weeks

This is a small, clean campaign, and that is why it is useful. One advertiser, one product, one format, a flat bid, four weeks, every conversion tracked by postback. Nothing was optimized automatically, so the numbers show what Telegram Mini App traffic does on its own.

The product is an Android video app published on Google Play. The advertiser promoted it with short video creatives inside partner Telegram Mini Apps, in the United States and Germany.


The setup

SettingValue
FormatTelegram Mini Apps
Creatives3 short videos per campaign
GeosUnited States (one campaign), Germany (one campaign)
BidFlat CPM $3.20, no Auto CPA
Frequency cap2 impressions per user per day
Daily budgets$30 (US), $20 (DE)
Conversion eventDefined by the advertiser, sent by S2S postback with our click ID
PeriodOctober 1–28, 2026 (UTC)

The conversion is the advertiser’s own event. We do not see what the app counts — install, first launch or sign-up — only that the postback arrives, on median about a minute after the click. Treat the CPA below as “cost per advertiser-defined conversion”, not as cost per paying user.

The result

CampaignSpendImpressionsClicksCTRConversionsClick → conversionCPCCPA
United States$300.3793,86712,33813.14%2361.91%$0.024$1.27
Germany$180.8656,5209,36216.56%1481.58%$0.019$1.22
Total$481.24150,38721,70014.43%3841.77%$0.022$1.25

Dashboard views below show the campaigns’ production numbers for October 1–28, 2026 in the advertiser Analytics page; campaign names and the account balance are blurred.

Advertiser dashboard, Analytics by campaign: October 1–28, 2026, campaign names blurred Advertiser dashboard, Analytics by campaign: October 1–28, 2026, campaign names blurred.

Two things stand out. The click-through rate is 14%: a video ad inside a Mini App gets attention that a banner on a web page does not. And the two countries land within five cents of each other on CPA, even though the US campaign pays more per click: US users who clicked converted more often.

Analytics by country: United States and Germany, October 1–28, 2026 Analytics by country: United States and Germany, October 1–28, 2026.

Week by week: what actually moves CPA

Spend grew more than tenfold over the month with the bid unchanged at $3.20 CPM.

Analytics by date: daily impressions over the month. The dips are days with little or no delivery: October 11–14 and October 19 Analytics by date: daily impressions over the month. The dips are days with little or no delivery: October 11–14 and October 19.

Week (from)SpendImpressionsClicksCTRConversionsClick → conversionCPA
Oct 1*$20.596,4332,07432.2%351.69%$0.59
Oct 5$39.9212,4743,52928.3%701.98%$0.57
Oct 12$72.6222,6944,79721.1%821.71%$0.89
Oct 19$217.6468,0147,84711.5%1321.68%$1.65
Oct 26*$130.4740,7723,4538.5%651.88%$2.01

* partial weeks: October 1–4 and October 26–28.

CPA went from $0.59 to about $2, and it would be easy to read that as “the traffic got worse”. It did not. Conversion rate per click stayed between 1.7% and 2.0% every single week. What fell was the click-through rate, from 32% to under 9%.

At a fixed CPM the arithmetic is simple:

CPA = CPM ÷ (1,000 × CTR × conversion rate)

With CPM and conversion rate flat, CPA moves inversely with CTR. The first weeks ran on a few placements where the video got clicked a lot; the extra volume came from a wider set of Mini Apps where users clicked less often but, once they clicked, converted just as well.

Where the conversions came from

  • 145 placements delivered clicks during the month; 34 of them produced at least one conversion.
  • The largest placement brought 19.7% of the clicks. No single source carried the result.

That spread is what makes the numbers trustworthy: 384 conversions from 34 sources is a pattern, not one lucky placement.

What to take from this

  1. Track the event you pay for. This advertiser’s postback is the only reason the table has a CPA column at all.
  2. Before you call a source bad, split CPA into CTR and conversion rate. Here the users were as good in week four as in week one; only the price per click changed.
  3. Video works in Mini Apps. A 14% blended CTR is what makes a $3.20 CPM land at two cents per click.
  4. Scale with a cap on CPA, not on CPM. The next step for a campaign like this is to keep the converting placements and let Auto CPA bid them toward a target, instead of buying every new impression at the same flat price.

Numbers are pulled from Adexium’s production data (impressions are sampled and weighted; clicks and conversions are counted one by one). This is one campaign’s outcome, not a typical result or a performance guarantee.

More about Adexium

Frequently asked questions

What is the minimum deposit?
Minimum deposit is $100 for advertisers. Funds are credited within minutes for crypto and instantly for cards.
What payment methods are supported?
We accept a wide range of payment options to fit any geography:
USDT
USDTTRC-20, ERC-20, BEP-20, TON
TON
TONToncoin native
Bitcoin
BitcoinBTC native
Ethereum Ethereum
EthereumETH + ERC-20
Capitalist
CapitalistUSD / EUR
Visa / Mastercard Visa / Mastercard
Visa / MastercardInstant top-up
Wire transfer
Wire transferSWIFT / SEPA
Advertisers: top up from $100. Publishers: payouts every 3 business days, min $25.
What geos are supported?
225+ geos worldwide, including tier-1 (United States, United Kingdom, Germany, France, Canada, Australia), tier-2 (Eastern Europe, LATAM, SEA), and tier-3 markets. Targeting works down to city level.
How do payouts work for publishers?
Automated payouts every 3 business days. Minimum payout is $25. We pay in USDT on TON, TRC-20, BEP-20 and ERC-20. On withdrawals of $50 or more we cover the network fee, so you receive the full amount; below $50 the network fee is deducted from the payout ($3 on TRC-20, $1 on TON, BEP-20 and ERC-20). No manual approvals, no hidden fees.
What tracking and postbacks are supported?
Server-to-server (S2S) postbacks, pixel tracking, and macros for all major trackers: Voluum, Keitaro, Binom, RedTrack, BeMob. Real-time conversion reporting.
How does anti-fraud work?
Three layers, none of them a black box. Before the auction, the bidder rejects requests from iframes, datacenter ASNs, hammering IPs and malformed user-agents; proxy/VPN traffic is blocked on SSP supply and targetable by a flag on direct placements. On the click, a JavaScript challenge by Kaminari — an independent antifraud vendor — filters bots before the redirect. Daily, every placement is scored on outcomes: enough clicks with zero conversions for any advertiser, or the same email registering across accounts, and it is blocked network-wide. Confirmed fraud is refunded.
Can I whitelist or blacklist sources?
Yes — whitelist and blacklist by source ID, app ID and zone, per campaign. There is no optimizer that silently drops sources for you. If you want automation, Auto CPA re-bids per source toward your target cost per conversion and leaves the source list under your control.
Can I pay per conversion instead of CPM/CPC?
Yes, two ways. Auto CPA bidding is self-serve on every format: you set a target cost per conversion and the bidder tunes CPM/CPC bids per traffic source to hit it. CPA pricing, where you pay a fixed price per conversion and nothing otherwise, is in beta and available on request — it runs on direct-publisher In-Page Push and Popunder inventory only, so we agree the conversion event and geos with you first. Both require S2S postback conversion tracking.