Performance Marketing
2026-10-17 · 10 min read
Sapun Lamichhane
Founder & CEO of Arcetis
Anyone offering a single "correct" Facebook ads budget for every Nepali business is skipping the actual variables that determine the right number: what the goal is (brand awareness versus direct leads versus online sales), what margin exists on whatever is being sold, and how much the business can tolerate spending during the inevitable testing period before anything is optimized. A clothing brand with 40% margins and a NPR 1,500 average order value has a completely different sensible budget than a custom furniture maker selling a handful of large-ticket pieces a month.
What is consistent across businesses is the structure of the decision: start with a number small enough that losing it during testing would not hurt, learn what works, and then scale the budget in proportion to what is actually converting - not the other way around.
It also helps to separate the budget question from the creative question - a business can spend a large amount with weak creative and underperform a smaller budget paired with a genuinely compelling offer and clear visuals, which is why the two decisions should never be made in isolation from each other.
“The question is not "how much should I spend on Facebook ads" - it is "how much can I afford to spend finding out what works before anything starts paying for itself."”
Meta ad costs in Nepal vary by audience and competition, but a cost per click commonly falls somewhere in the NPR 5-25 range for a reasonably targeted campaign, and cost per lead (a person submitting contact information) often lands anywhere from NPR 50 to a few hundred rupees depending on how specific and high-intent the offer is. These are directional, not guarantees - a highly competitive niche or a poorly targeted campaign can run well outside these ranges in either direction.
A commonly cited rule of thumb for getting a campaign out of Meta's initial "learning phase" - where the algorithm is still figuring out who to show ads to - is roughly NPR 500-1,000 per day sustained for at least a few days to a week. Below that, campaigns often never gather enough data to optimize properly, which is a common reason small test budgets "don't work" when the real issue is that they never left the learning phase.
These figures also move with the calendar - costs in Nepal tend to rise around major shopping periods and festivals such as Dashain and Tihar, when many more businesses are competing for the same audience's attention at once, and a budget that performed well in an ordinary month may need adjusting to maintain the same results during these higher-competition windows.
For an established business with existing revenue, a commonly used starting benchmark is allocating somewhere between 5% and 15% of revenue toward paid advertising, adjusted up for a business in growth mode trying to acquire market share quickly, or down for a business that is margin-constrained or primarily relies on repeat customers and referrals. A brand-new business with no revenue yet obviously cannot use this method and instead needs to budget from available capital, treating early ad spend explicitly as a cost of market research.
Management fees are a separate line item from ad spend and should never be confused with it - in Nepal, management for a single ad channel like Meta typically runs NPR 15,000-30,000 per month, independent of however much money is actually being spent on the ads themselves.
It is also reasonable to set aside a small, separate testing budget on an ongoing basis even after a campaign is performing well, since audiences and creative fatigue over time - a winning ad from three months ago will typically start underperforming eventually, and a business that stops testing new variations once something works tends to see gradual decline rather than sudden failure, which makes it easy to miss until it has already cost real money.
A small local service business (a salon, a tutoring center) testing Facebook ads for the first time might reasonably start with NPR 15,000-20,000 per month in ad spend, enough to clear the learning phase and gather a few weeks of real data without betting the business on it. A growing e-commerce brand with proven unit economics might run NPR 80,000-200,000+ per month once it has a campaign that reliably returns more than it costs. A business purely focused on brand awareness ahead of a launch, where conversions are not the immediate goal, needs a different budget logic entirely, usually built around reach and frequency targets rather than cost per lead.
None of these numbers are prescriptive for any specific business - they are reference points to sanity-check a budget against, not a formula to copy exactly.
A fourth scenario worth naming directly: a seasonal business, such as one tied to trekking season or a specific festival period, needs a budget that concentrates spend into the weeks that actually matter rather than spreading it evenly across the year, which is a common mistake for businesses applying a generic monthly budgeting habit to an inherently seasonal demand curve.
A business that spends NPR 20,000 well - with clear targeting, a specific offer, and proper conversion tracking - will consistently outperform a business that spends NPR 100,000 poorly, with vague targeting and no way to tell which ad actually drove a sale. Before increasing budget, it is worth confirming the basics are in place: a pixel or conversion API tracking actual results, creative that has been tested in more than one version, and an honest sense of what a profitable cost per lead or sale actually looks like for that specific business's margins.
Arcetis has managed more than $70,000 USD in Meta ad spend across client accounts and generally advises starting smaller than instinct suggests, proving the mechanics work, and only then scaling the number - rather than committing a large budget upfront and hoping the targeting is right from day one.
Before increasing spend on a campaign that already looks to be working, it is also worth checking whether growth is coming from reaching new people or simply showing more ads to the same people more often - the latter tends to produce diminishing returns much faster than the former, and distinguishing between the two is one of the more overlooked diagnostics available directly inside the ad platform's own reporting.
Roughly NPR 500-1,000 per day sustained for at least a week is generally enough to get a campaign through Meta's learning phase and gather usable data, though the exact figure depends on the industry and how specific the targeting is.
A common starting benchmark for an established business is 5-15% of revenue, adjusted based on margin and growth goals. A pre-revenue business should instead budget from available capital and treat early spend as market research.
No, and conflating the two is a common and costly mistake. Management fees for running a single channel like Meta typically run NPR 15,000-30,000 per month in Nepal, entirely separate from whatever is actually spent on the ads themselves.