RECORD01/04
Joytime, Vietnam
Joytime
Sole owner of a USD 20,000 paid media budget across four platforms, where most of the work was proving which half of the spend could not be defended.
- Role
- Paid media, freelance
- Engagement
- Jan to Jun 2026
- Channels
- Meta Ads, TikTok Ads, Google Ads, Apple Search Ads
READOUT02/04
Measured outcomes
USD 20,000
paid media managed
Across four platforms and six markets in six months.
MEASURED
57.8%
of social budget flagged
Sitting in objectives that could not be defended on performance.
FLAGGED
63×
gap in cost per site visit
Between the cheapest and the most expensive objective in the same account.
FLAGGED
924,874
conversions that were not real
A misfiring TikTok event reported them at VND 109 to 203 each, against VND 25,046 for the event that was.
FLAGGED
18×
lower cost per site visit
After the March rebuild of the account structure.
REDUCED
4,700
app installs delivered
At USD 0.80 each, across six markets.
GAINED
14×
the going rate per install
What two campaigns were paying before I cut them.
FLAGGED
GRAPHICS03/04
Data graphics
REDUCED
FLAGGED
ACCOUNT04/04
The situation
Joytime brought me in as the only person on paid media. Six months, roughly USD 20,000, four platforms, six markets, and no one else to hand a question to.
The account was already running when I arrived and it was reporting well. That was the problem. Objectives had been chosen for the metrics they were good at producing, so the dashboard looked healthy while the money underneath it did very little.
What I did
- Audited every active objective against a single question: what does one site visit cost here, and can I defend that number to someone paying for it.
- Found 57.8% of the social budget in objectives that failed that question, and a 63× spread in cost per site visit inside the same account.
- Traced a TikTok conversion event reporting 924,874 conversions at VND 109 to 203 each, against VND 25,046 for the event that mattered. The cheap one was counting something that was not a conversion.
- Rebuilt the account structure in March around the objectives that survived the audit, and rewired the GA4 events and UTMs so the next audit would take an afternoon.
What changed
Cost per site visit fell 18× after the rebuild. Nothing clever happened. Spend that was buying impressions and calling them outcomes got removed, and the average stopped carrying it.
The account delivered 4,700 app installs at USD 0.80 each over the six months. Two campaigns were paying fourteen times that rate for the same outcome.
The conversion event was the more useful catch. Left alone, it would have kept justifying its own budget forever, because a metric reporting 924,874 of anything at VND 109 each always wins an argument against one reporting fewer at VND 25,046.
What I would do differently
I audited the objectives before I audited the events, and it should have been the other way round. A broken event does not just waste its own budget, it corrupts every comparison downstream of it, including the audit I was in the middle of running.
I can show what a site visit cost and how much of the budget could not justify itself. I cannot show what any of it earned, because there was no revenue tracking in place while I was there.