How Much to Spend to Test Paid Ads in Saudi Arabia and the UAE

Ask how much you need to test paid ads in Riyadh or Dubai and you will usually get a round number: a few thousand dollars a month, more for LinkedIn. Those figures are rarely sourced, and they skip the question that matters. A test budget is not a fee for entering the auction. It is the amount of money needed for the platform’s algorithm to collect enough results to optimise, and for you to collect enough results to decide.

That amount depends on three things you can estimate before you spend: what the platform needs in order to learn, what one result is likely to cost you, and how many separate things you are trying to test at once. This guide shows how to put those together, then adds the local costs Saudi and UAE companies often leave out of the calculation.

Minimum budgets are not test budgets

Every platform publishes a floor below which it will not run a campaign. These floors are low, and they tell you almost nothing about whether a test will work.

• Meta. Its Marketing API documentation lists minimum daily budgets for ad sets on automatic bidding of USD 0.50 when you pay for impressions, USD 2.50 for clicks, and USD 40 for low-frequency actions, converted into your account currency. Meta labels these amounts as illustrative and says they can change; it doubles them for a listed group of higher-cost countries, which does not include Saudi Arabia or the UAE (Meta for Developers, ad set reference).
• TikTok. Its help centre says a campaign daily budget must exceed USD 50 and an ad group daily budget must exceed USD 20 (TikTok Ads help, Budget).
• LinkedIn and Snapchat also enforce minimum daily and total budgets in their ad managers. Check the current figure in your own account and currency before planning, because these change and differ by setting.

A campaign that meets the minimum will deliver ads. It will not necessarily deliver enough results for the bidding system to settle, which is the real cost of a test.

What each platform needs before its results mean anything

Automated bidding works by learning which people, placements and moments produce the result you asked for. Until it has seen enough of those results, delivery is unstable and early cost figures are unreliable.

Two practical consequences follow. First, the number of ad sets or campaigns you run multiplies the money needed, because each one learns separately. Second, the event you optimise for decides the budget: a purchase or a qualified lead is rarer and more expensive than a landing-page view, so it needs more spend to reach the same volume.

The test-budget formula, with a worked example

Use this sequence instead of a benchmark. Every input below is an assumption you replace with your own figures. The two examples are hypothetical businesses with made-up numbers, not client data or benchmarks.

1. Events needed per test unit per week (E). Take the platform’s current guidance for the event you optimise for. For a Meta ad set, the help centre’s figure of about 50 a week is the usual starting point.
2. Expected cost per event (C). Use your own history if you have it. If you do not, run a short, small pilot first to get a rough figure, rather than borrowing someone else’s benchmark.
3. Number of test units (N). Ad sets on Meta, campaigns on Google, ad groups on TikTok. Fewer is cheaper and cleaner.
4. Test length in weeks (W). Long enough to cover learning plus a decision window. On Google, at least two conversion cycles or two weeks.
5. Media budget = E x C x N x W. Then add the local costs in the next section.

Example A (hypothetical), Meta lead campaign. Imagine a Riyadh training academy that wants to test Meta lead ads for course inquiries, optimised for submitted leads. A small pilot suggests about SAR 60 per lead (C); it wants one ad set to exit learning (N = 1) and plans a four-week test (W = 4). Using about 50 events a week (E): 50 x 60 x 1 x 4 = SAR 12,000 in media, or roughly SAR 430 a day. If it splits the same test across three audiences, the figure triples. If it cannot fund that, it changes the design, not the expectations (see below).

Example B (hypothetical), Google Search with Smart Bidding. Imagine a Dubai services firm that wants to judge a Search campaign on at least 30 conversions. It assumes 5% of clicks become an inquiry and an average click cost of AED 6. It needs 30 / 0.05 = 600 clicks, so 600 x 6 = AED 3,600 in media over at least two weeks, assuming search demand exists to supply 600 clicks in that time. If the keyword volume is too thin, the test will take longer regardless of budget.

Two checks keep the maths honest. Google can spend up to twice your average daily budget on a given day, while spending no more than 30.4 times the average daily budget over a month, so plan cash flow on the monthly figure, not the daily one. And a cost per event that looks fine is only useful if those events turn into sales conversations, so decide in advance how you will count qualified inquiries. Our guide to setting up GA4 key events for lead tracking covers that step.

The costs the formula misses in Saudi Arabia and the UAE

VAT on ad invoices

Ad platforms often bill Gulf advertisers from entities outside the country. Check the billing entity on your invoice, because it changes how VAT works.

• Saudi Arabia (15%). ZATCA’s guideline on imports and exports (second edition, May 2026) explains that when a non-resident supplies services to a VAT-registered business in the Kingdom, the Saudi recipient accounts for the VAT under the reverse charge mechanism, declaring output tax and, if entitled to full deduction, deducting the same amount as input tax. If the recipient is not VAT-registered, the reverse charge does not apply and the non-resident supplier is responsible for the tax. The same guideline notes that services received from abroad count towards the SAR 375,000 mandatory registration threshold (ZATCA guideline).
• UAE (5%). A VAT-registered UAE business that receives services from a supplier abroad generally accounts for the VAT itself under the reverse charge, as summarised in Andersen’s note on FTA public clarification VATP044 (2025). For Meta specifically, the most detailed public report we found is older: when Facebook began charging VAT on UAE ads in 2019, Gulf News reported that advertisers without a valid 15-digit tax registration number (TRN) in their account would be charged 5%, and that Facebook verifies the number with the UAE tax authority. Billing set-ups may have changed since then, so check your current invoices and account tax settings.

In practice: if your company is VAT-registered, add the tax number to every ad account and make sure your accountant records the reverse charge. If it is not registered, budget for the VAT on top of media. Rules and platform billing entities change, so check the current invoice wording and confirm treatment with your tax adviser. This is not tax advice.

Currency and payment

Check which currency each ad account bills in and how your company pays. A card in riyals or dirhams paying a dollar-billed account can attract foreign-transaction fees from your bank, which belong in the test budget. Set the account currency deliberately when you create it rather than accepting a default.

The non-media costs

• Tracking. Pixel or Conversions API setup, Google Ads conversion actions and a way to record WhatsApp or call inquiries. Without these, a test produces spend but no answer.
• Creative. At least two or three versions per test unit, in the language your buyer actually reads, Arabic, English or both.
• Landing path. A focused page or a WhatsApp flow that answers the questions an ad raises (see landing page or homepage for where the traffic should land).
• Sales time. Someone has to call every lead quickly. Leads that wait are part of the cost and can make a sound channel look like it failed.

When the formula says you cannot afford the test

This is a common result, especially for B2B companies with expensive leads. Adjust the design in this order before abandoning the test:

1. Test fewer things. One ad set or campaign, one audience, one offer. Splitting a small budget across five audiences guarantees none of them learns.
2. Start where intent already exists. If buyers search for what you sell, Google Search on a tight set of keywords often needs less money to produce a decision than prospecting on social. The Saudi platform split guide covers matching platforms to buyer types.
3. Optimise for a more frequent event, carefully. A WhatsApp conversation started or a landing-page view happens more often than a submitted quotation request, so the system learns faster. The trade-off is that it will find people who do that action, not necessarily people who buy. Judge the test on qualified inquiries even if the bid optimises for something earlier; our diagnosis of lead form ads that bring fake numbers shows what happens when that check is skipped.
4. Extend the timeline. A lower daily budget over more weeks can still reach a decision, though the ad set may stay in learning longer and costs will be noisier.
5. Narrow the geography only if sales does. Testing one city makes sense when that is where you can deliver and follow up; otherwise it just shrinks the audience.

A test plan you can hand to your team or agency

Write these down before the first riyal or dirham is spent:

• The question. For example: can Meta lead ads produce qualified inquiries for our maintenance contracts below an acceptable cost?
• The success measure and its threshold. Cost per qualified inquiry, not cost per click or per lead form, and the maximum you can pay given your average deal value and close rate.
• The budget calculation using the formula above, with VAT, fees and creative listed separately.
• The no-touch window. No significant edits during learning unless something is broken.
• The decision date and what happens at each outcome: scale, adjust and rerun, or stop.
• Ownership. Ad accounts, pixels and audiences sit in the company’s name so the results stay with the business whoever runs the campaigns.
• Season. Note whether the test overlaps Ramadan, Eid, a national-day period or a major sale, and do not compare it directly with a test run outside that period.

RX plans performance marketing tests the same way: the question and the success threshold are agreed first, ad accounts stay in the client’s name, and results are judged on qualified inquiries. When the harder question is whether paid ads are the right channel at all, that belongs in a marketing plan before the test is designed.