There is no single marketing budget that works for every Hong Kong business. The useful starting point is what a new customer contributes, how long it takes to get paid and how much uncertainty the business can afford.
This is a planning framework, not a survey of Hong Kong spending. Any numbers below are explicitly hypothetical calculations, not CLS client results or market benchmarks.
Separate the full cost from the media budget
List media spend, agency or staff cost, production, website work and tools. Decide which costs belong to a particular test and which support several campaigns. Comparing an all-in quote with an ad-management-only quote is not a like-for-like comparison.
For CLS's own terms, use the current pricing page: HK$20,000–40,000 per month for the agreed ongoing scope, with ad spend and applicable additional costs outlined separately. That is our offer, not an industry minimum.
Work backwards from contribution
For an online sale, begin with revenue after refunds and discounts. Subtract product cost, fulfilment, payment fees and other variable costs to estimate contribution before marketing. The remaining amount helps establish an acquisition ceiling; it is not all available to spend if you also need to cover overhead and profit.
Hypothetical example: a HK$1,000 order with HK$600 of variable cost leaves HK$400 before marketing and fixed overhead. Spending HK$300 to acquire that order leaves HK$100 before fixed overhead. A revenue-based ROAS of 3.33× does not by itself establish profitability.
For a lead-driven service, use the probability that an enquiry becomes a paying customer. If an acquired customer contributes HK$2,000 before marketing and one in ten enquiries becomes a customer, HK$200 per enquiry would consume that entire contribution. A viable target would need room for sales cost, overhead, risk and profit.
Choose a test you can actually interpret
Agree the audience, offer, conversion definition and decision window. Estimate likely traffic and outcomes from your own records where available. With no history, set a spending cap and treat the first test as uncertain rather than presenting a forecast as fact.
A small budget spread across many channels may leave each test inconclusive. Prioritise the strongest customer route first. Search may suit an existing demand; social creative may explain an unfamiliar offer. The right allocation depends on the business, not a universal percentage split.
Define the stop, revise and expand decisions
- Stop: broken tracking, unavailable stock, an unusable form or costs beyond the agreed risk limit.
- Revise: relevant traffic but poor enquiry quality, weak follow-up or a mismatch between the advertisement and page.
- Consider expansion: repeatable qualified outcomes within the business's economics, with capacity to serve more customers.
More spend does not guarantee proportional growth. Review the marginal cost of additional customers as well as the campaign average. Allow for sales delays and refunds before treating a short-term result as settled.
Keep revenue reporting honest
ROAS compares attributed revenue with advertising spend; it is not an all-in profit measure. Do not add revenue claimed by two platforms without checking overlapping attribution. Reconcile orders or completed deals against the source business records.
Our marketing measurement framework explains the definitions. For implementation, Google's conversion guidance distinguishes actions such as purchases, calls and offline outcomes. Pick the action that represents your actual goal.
What should happen in the first review?
Review spend against the cap, tracking quality, the number and quality of outcomes, and what needs to change. SEO, new offers and long sales cycles may need longer observation. A 90-day plan is a work period, not a guarantee of positive ROI.
For a proposed allocation, bring your margins, recent sales, existing marketing costs and capacity constraints. Our advertising service connects channel decisions to creative and measurement within that context.
Editorial review: 30 September 2026. Practical guidance; examples are not performance guarantees.
