The Ad That Won the Most Installs Lost on Customers: UGLYCASH on TikTok

By Kevin Veitia

Client

UGLYCASH, a US dollar account and card for Latin America

Channels

TikTok Ads, with Meta Ads for comparison

Timeframe

March to April 2025

Goal

Cost per first funding

Measurement

TikTok Ads Manager and the app's attribution tool

The challenge

UGLYCASH gives people in Latin America a US dollar account and a Visa card. We covered the wider engagement, seven markets on three platforms, in the main UGLYCASH case study. This one is about TikTok, the channel that looked best on the surface and worst underneath.

By early March 2025, TikTok's click-through rates were around 2 to 2.5%, well ahead of what Meta was getting. But roughly $1,000 a day was producing only a handful of visible first fundings, the moment a verified customer actually puts money into the account. For a fintech app, that's the only number that pays the bills.

Fixing the structure first

Before judging any ad, we fixed the structure. The account had been split into too many narrow campaigns, all chasing the same conversion events. They competed with each other for the same people, and none of them gathered enough conversions to get out of TikTok's learning phase, the period when the algorithm is still working out who to show ads to. We consolidated them, set clear rules for when a campaign had earned its way out of learning, and took budget out of the markets that weren't pulling their weight.

By 17 April, TikTok was producing first fundings at about $90 each on Android and $111 on iOS. At those costs, we estimated it could deliver about 320 first fundings a month on iOS at $35,000 a month in spend, as long as creative production could keep up with ad fatigue.

That fixed the cost. It didn't answer the more useful question: which ads were bringing in people who would actually fund?

Reading every script

TikTok creative wears out in days, so testing has to be fast. We pulled the results for every ad, from install through identity verification (KYC) to first funding, matched them to each script, and used AI-assisted analysis to compare the scripts and surface the patterns quickly.

One note on the numbers. The install counts below come from the app's attribution tool, which counts far more installs than TikTok's own dashboard does. Every comparison stays within that one source.

What the data showed

The best install ad was the worst at customers

Two ads drove almost the same number of installs. "11% Rewards", built on a headline rewards rate, brought in 10,979 installs and 7 first fundings: 0.06%. "Virtual Account vs PayPal", which told people they could get paid from PayPal, Payoneer and Deel with no fees, brought in 10,641 installs and 36 first fundings. Same volume at the top, five times the customers at the bottom.

The rewards rate attracted people browsing for a deal. The payment message attracted people with a job to do: receiving money from abroad, like freelancers paid through Payoneer or Deel. They had a reason to fund the account on day one.

Creators beat brand ads where trust matters

Creator videos brought in fewer installs, but far better ones. A travel creator's video, pricing out a trip to Turkey and mentioning the card's 1% cashback along the way, turned 16.2% of its installs into verified accounts, against 1.6% for the rewards ad. Another creator's video took about 2% of its installs all the way to first funding, roughly 30 times the rewards ad's rate, though on a small sample of 6 fundings.

Verification means handing over your ID, and funding means handing over your money. A real person using the card on a real trip did what no headline rate could: it made the product feel safe enough to do both.

The bigger the ask, the shorter the script

On TikTok, the top install ads were one or two sentences, under 100 characters. Meta told the same story from a different angle. A long, rewards-heavy video script won the cheapest installs there, at $3.28 each. But a 398-character script about getting a card issued abroad delivered 73% of all Meta video identity checks, at $7.18 each. The more an action asks of someone, the less the ad should say: one tangible benefit, and a reason to act now.

What we recommended next

• Judge every ad on cost per first funding, never on installs or click-through rate
• Lead funding campaigns with the payment message, naming the platforms people already use
• Put more budget behind creator videos that show the card in real use, like travel and online shopping
• Keep install scripts short and benefit-first, and give verification campaigns one tangible benefit, like the physical card
• Test new ads in a separate campaign and move only the winners into the scaling campaign, so weak ads don't feed the algorithm bad data

Results

• TikTok cost per first funding down to about $90 on Android and $111 on iOS by 17 April 2025, from roughly $1,000 a day for a handful of fundings in early March
• Five times the first fundings from the payment message as from the rewards message, on almost the same installs
• Creator videos verifying installs at up to ten times the rate of the rewards ad
• An estimate of about 320 first fundings a month on iOS at $35,000 a month, if creative production could keep pace

Some of these samples are small: 36 fundings for the best brand ad, 6 for the best creator video. They're directional, not final. But the same pattern showed up on two platforms and at two steps of the funnel, which is more than any single number can tell you.

UGLYCASH came back to us in 2026, when we built a measurement and experimentation framework for its paid acquisition, starting with Meta.

What you can take from this

• Judge creative on the customer, not the click. The ad with the most installs can be the worst at producing funded accounts.
• Sell the job, not the rate. People with a reason to use the product today convert; people chasing a number browse.
• In fintech, creators earn trust that brand ads can't. Use them for the steps that ask for an ID or a deposit.
• The bigger the ask, the shorter the script.
• Consolidate before you judge the creative. Over-segmented campaigns split the same conversions until none of them gets out of learning.
• AI makes fast creative analysis practical. Feed it funnel data, not dashboard metrics.

Want a second set of eyes on your account? Book a free 30-minute consultation and we'll walk through what's working, what isn't, and where the next win is. Curious what it costs first? Our pricing is public.