I decide where the money goes, then I go and spend it.
I run around $180K a month in paid media across Meta and Google. hoichoi is one of the accounts. Most of the work happens before anyone opens an ad account.
The allocation call comes first
Before any channel work there is one decision. How much goes out this month, where it goes, and what it is expected to bring back. I run that through budget planning and revenue projections, then hold the plan against actuals every week and every month.
At that scale it is roughly $180K a month, close to $2M a year, split across Meta and Google and across a core market and international ones. The channel split is an output of the plan, not the starting point.
Reallocation is where most of the value sits. Moving budget between markets, between prospecting and reactivation, between platforms, based on what the last fortnight actually returned rather than what the last plan assumed.
Running Meta and Google, and the reporting underneath
I run execution end to end. Structure, budgets, bids, audiences, launch, and the daily read on whether anything needs to change.
Reporting on spend against targets goes out weekly and monthly, to the team and to leadership. That cadence is only worth anything if the numbers are trustworthy, so I own the tracking it sits on: events, CAPI, OCI, conversion actions and MMP data, tied back to the payment gateways.
Keeping tracking and reporting in the same pair of hands is deliberate. When a number looks wrong I go and find out why instead of filing a ticket and waiting a week to be told it was a mapping issue.
Testing, and reading what comes back
I have run the experiment and creative testing setup for a large paid acquisition and reactivation programme spanning core and international markets. Cohorts, holdouts, and a schedule that stops tests from colliding with each other.
Creative is where most of the remaining headroom in an account sits, so the testing has to be able to isolate it. A promo ID on every asset, consistent nomenclature in the account, and results that tie back to the specific creative that earned them rather than to a campaign average.
That work helped the business reach its highest revenue month while spending at a record high. Pushing spend to a record and still having return follow is the part the testing discipline pays for.
Zero to one, and the work before a channel exists
I took direct carrier billing in international pockets from nothing to a working business channel at hoichoi. No baseline, no account history, no existing creative, no internal precedent for how it should be measured. I have since run the same kind of build in microdrama with Sooper.
On the finance side I worked on GTM with the co-founder and CEO at StockEdge and elearnmarkets, including a new wealth management business. Market research, positioning, launch planning, pitch decks. I also owned ASO and UAC, where store visibility and keyword work moved alongside the paid side rather than on a separate track from it.
Through SVF Brands I have planned and run performance for MPJ Jewellers and Priya Gopal Bishoyi. D2C retail behaves differently from OTT. Catalogue, margin and purchase cycle set the ceiling, and the media plan has to respect that rather than argue with it.
Where this stops working
Paid acquisition is an amplifier. If pricing, catalogue or onboarding is wrong, spend finds that ceiling fast and no amount of account structure moves it. I have spent weeks on channel work that a product change would have solved in a day.
The allocation method depends on attribution being sound. Where tracking is thin, or where the payment path does not report back cleanly, every reallocation is a guess. In that situation I fix the measurement first and hold the budget flat.
It is also overbuilt for small budgets. Under a certain monthly spend, the testing apparatus costs more time than the decisions it informs are worth. Run two things, look at what came back, and skip the infrastructure until there is enough money moving to justify it.
