Your first ad budget as an SME: how to split it without burning money

The most common question from a small business owner running ads for the first time is not "how do I write the ad". It is "how much money do I put in, and how do I split it so it does not vanish in two weeks". This article answers exactly that, with concrete numbers for the Vietnamese market.

How does the 70/20/10 rule work?

First time advertisers usually fall into one of two extremes: either they pour everything into a single ad and hope, or they split the money across 5 to 7 ad sets to "see what happens". Both waste money. The first has no fallback when the guess is wrong. The second means no ad set ever accumulates enough data to conclude anything, a problem covered in depth in measuring ad performance: which metrics actually matter.

The 70/20/10 rule sits between those extremes. 70 percent of the budget goes to the core: one or two ad sets on the channel and audience you have the strongest reason to believe in, for example a lookalike of past buyers, or the search keywords that describe your service exactly. 20 percent goes to variations of that core: same audience with different creative, or the same creative on an adjacent audience. 10 percent goes to entirely new ideas: an untried channel or an untried format, for example short video instead of static images.

The important part is the monthly loop: at the end of each cycle, any variation or new idea that beats the core gets promoted to receive next month's 70 percent. Budget keeps flowing toward whatever has the best data, instead of whatever you personally like most.

What is the minimum meaningful budget in Vietnam?

Good news first: Vietnam is one of the cheapest advertising markets in the region. According to ADCostly data (2025), the average cost per click for Facebook ads in Vietnam is around 0.16 USD, roughly 4,000 VND, and the average cost per 1,000 impressions is about 1.83 USD, roughly 47,000 VND. On Google, WordStream's cost per click data by country puts Vietnam about 76 percent below the US average, and local rate cards for 2026 list typical search click prices from 3,000 to 15,000 VND depending on the industry. Statista forecasts the whole Vietnamese digital advertising market at around 1.45 billion USD for 2025, with search advertising as the largest segment at roughly 623 million USD.

Cheap does not mean any amount works. A "meaningful" budget is one that buys enough data to conclude something. Work backwards: say you need 30 enquiries to compare two ad sets, and your landing page converts at 2 percent, then you need about 1,500 clicks. At 4,000 VND per click, that is 6 million VND. So a practical starting point for the first test month is 6 to 10 million VND, about 230 to 380 USD, concentrated on a single platform, which works out to 200,000 to 330,000 VND per day.

The lower bound needs saying plainly too: below 3 million VND a month split across two platforms, each ad set gets a few tens of thousands of VND a day and the data that comes back is almost pure noise. In that case, saving for another month or two and then running one proper test beats dripping money all year.

How long should a test run before you conclude?

Two weeks is the minimum, and the reason is not a gut feeling. First, every ad platform has a machine learning warm up phase: Meta's delivery system recommends around 50 conversions per ad set within a week to exit its learning phase. Results inside that phase swing widely and do not reflect true performance. Second, buying behaviour differs a lot by day of the week; running from Wednesday to Saturday and then concluding means reading a fragment of the picture.

A lean schedule: in weeks 1 and 2, do not touch the campaign structure at all, just observe. At the end of week 2, compare ad sets by cost per enquiry, not by click price. In weeks 3 and 4, gradually shift money to the leader and replace creative on the laggards. If your product has a long consideration cycle, for example customers ask for a price and close 10 days later, add exactly that cycle to the test window before judging anything.

When should you switch a campaign off?

Switch off too early and you kill a good campaign by mistake; too late and you burn money. Three practical thresholds, written down before you press start:

• Spent 3 times the target cost with nothing to show: if you are willing to pay 100,000 VND per enquiry, an ad set that has spent 300,000 VND without a single enquiry gets paused, no need to wait out the week.
• Enough data, still losing: when an ad set has a few dozen conversions but cost per paying customer is still above gross profit per customer, and you have already tried new creative and a new landing page, switch it off and move the money.
• Creative saturation: frequency above 3 to 4 impressions per person with CTR falling week after week means the audience has gone numb to the creative. Replace the creative first; that does not necessarily mean killing the whole campaign.

The reverse warning also applies: do not switch a campaign off just because click prices rise slightly with the season. Click price is an operational metric; on and off decisions rest on cost per real customer.

What are the four most common money burning mistakes?

First, targeting too broad or too narrow. Too broad, for example "nationwide, ages 18 to 65", and the money flows to people who will never buy while cheap clicks trick you into thinking the campaign is working. Too narrow, for example an audience of a few tens of thousands, and impression prices climb, frequency spikes and the algorithm has no room to optimise. For cafes and shops serving a neighbourhood, a radius of 5 to 10 km around the store is a sensible starting point, in line with the approach in phygital for coffee shops.

Second, changing creative and budgets constantly. Every major edit makes the delivery system relearn from scratch, and the results for the next few days are close to meaningless. Beginners fall into a spiral: no orders by day 2 so they swap the image, day 4 the headline, day 6 the audience, and after two weeks nobody knows what is actually being tested. The discipline is one variable per change, and enough data before the next change.

Third, no proper landing page. Many campaigns point straight at a homepage or a social profile, where the visitor has no idea where to click to ask for a price. Ad money buys attention; the landing page is where attention turns into messages. A good landing page needs one message, one clear price or offer, and one contact button. Landing page copy and live selling share the same structure; see the live selling script framework for how attention is walked to a close.

Fourth, scaling the budget too fast. Seeing profit and tripling the budget overnight is the fastest way to break a working campaign, because the system must immediately expand into lower quality audiences. A safe increase is usually 20 to 30 percent at a time, spaced a few days apart.

One last leak sits outside the ads manager entirely: a customer messages at 9 pm and gets a reply at 8 am the next day. The click was paid for, the customer has already moved on. Before you raise any budget, make sure messages get answered within minutes, including out of hours; that is what an AI Agent on duty 24/7 does better than any human shift.

FAQ - Frequently asked questions