Deposit cost started sky-highhow Facebook Ads in Azerbaijan and Uzbekistan reached 78% ROAS by day 28

This campaign did not begin as a success story. Deposit cost was sky-high. Registrations were coming in and some first deposits appeared, but the economics did not work. Once we brought CPA FTD down, the next problem surfaced: too many players were depositing only the minimum amount.

Bringing in a deposit was not enough. We needed a valuable player, repeat deposits and enough time to see what the cohort looked like by day 28.

For four months, we ran Facebook Ads across Azerbaijan and Uzbekistan: testing, cutting weak combinations, relaunching the ones with potential and checking everything against deposits. CPA FTD eventually reached the target range and dropped below it in some stable combinations. ROAS reached 78% by day 28.

First, a note about the two reports

There are two data sources and they should not be mixed:

• Keitaro covers Uzbekistan traffic only;
• the affiliate platform combines Azerbaijan and Uzbekistan.

We do not add affiliate registrations to Keitaro conversions and pretend they form one perfect funnel. The reports cover different parts of the project, but together they show the scale and what happened after the first deposit.

What Keitaro shows for Uzbekistan

The Uzbekistan slice recorded 85,072 clicks, 6,294 conversions and 958 sales in tracker terminology. Sales represented 1.13% of clicks and 15.22% of conversions.

Then comes the more interesting part: 1,263 rebills and $13,735.21 in rebill income. Keitaro separately shows $9,617.15 in confirmed income.

Expense is zero because Facebook spend was not passed into the tracker. The 0% ROI on screen is therefore not the campaign result; it is an empty expense field. A dashboard can display a percentage, but without spend it says nothing about performance.

What the affiliate platform shows across AZ + UZ

Across both GEOs, the affiliate platform recorded 19,580 registrations and 2,807 depositing players. Registration-to-depositor conversion was 14.34%.

Those players made 7,195 deposits worth $197,426.91. On average, that means:

• 2.56 deposits per depositing player;
• $27.44 per deposit;
• $70.33 in total deposited value per depositor.

Reported income was $35,514.25, or $12.65 per player.

The key number is not only 2,807 players. It is the 7,195 deposits. Players did not stop after one action, and a meaningful share of value arrived later in the cohort lifecycle.

Why deposit cost went through the roof at first

When CPA FTD is high, it is tempting to change everything at once: creative, audience, app, landing page and offer. More money gets spent, while the team still has no idea what actually failed.

We broke the problem into stages:

1. Is creative attracting the right person or just cheap clicks?
2. Does the user reach registration?
3. Does registration turn into a first deposit?
4. How much does the player deposit?
5. Does the player return for a rebill?

The goal was not another cheaper lead. It was a combination that could survive the full journey to revenue.

How testing brought CPA FTD down

There was no magic button. The loop was constant: launch a hypothesis, collect data, check the lower funnel, kill the weak version, improve the strong one and run the next test.

Several rules kept the process under control:

• do not blend Azerbaijan and Uzbekistan creative conclusions;
• judge static and video by deposits, not likes or CTR;
• do not call cheap registration a win;
• do not scale until FTD quality is understood;
• wait for rebills and mature ROAS after the first deposit.

Multiple iterations first returned CPA FTD to a workable range, then pushed some stable combinations below target. It was not one genius banner. It was the buying, creative and analytics teams continuously sharing feedback.

Then came the minimum-deposit trap

Once FTD cost looked healthy, another issue became clear. Players were depositing, but too many were choosing the minimum amount. The tracker looked green because the event existed and volume grew. The business result was less exciting.

We stopped treating all FTDs as equal. Deposit size, the share of minimum deposits and rebills joined the control set. A combination was strong only when the players produced useful value after the first deposit.

At this stage, product feedback mattered more than another batch of creative. The algorithm sees the event it receives. The team must understand what that event is actually worth to the business.

Why one strong day was not a reason to celebrate

Conversion can swing hard in these GEOs. One day registration-to-deposit may run at 3:1; over the next few days it can move to 10:1. Reacting to every daily swing turns learning into chaos.

We therefore judged the cohort on day 28. ROAS had reached 78% by then. It was not a spectacular multiple from one lucky slice, but a cumulative result after repeat deposits had time to appear.

The exact formula and complete spend still need to be shown in one reconciled table, but the 78% result has been confirmed by the project team.

The four-month outcome

The project began with sky-high CPA FTD and too many minimum deposits. It ended with workable deposit cost, substantial repeat activity and 78% ROAS by day 28.

The combined two-GEO report shows 2,807 depositing players and 7,195 deposits worth $197,426.91. Keitaro separately shows 958 sales and 1,263 rebills for Uzbekistan.

Most importantly, this result did not come from one lucky launch. It was built through dozens of decisions: what to kill, what to relaunch, which creative to improve and which combination deserved time for the cohort to mature.

The next step is more volume, separated AZ and UZ product reporting, and another round of creative tests. Scaling will follow CPA FTD, deposit size and rebills, not one pretty day in Ads Manager.

Read our guide to acquiring depositing players, or explore iGaming marketing and Meta Ads management.