You cannot learn media buying from slides. The part that matters is the twenty minutes after a campaign goes live, when the numbers are not the numbers in the example, and the decision is whether to leave it alone. That requires a live account and a real budget, which raises two questions a syllabus rarely answers: how much, and who pays.
We are going to answer the first from platform documentation rather than from opinion, and be plain about the second.
What the platforms will actually let you spend
There is a documented floor, and it moved this year. On 27 February 2026 Google's Ads API team published a minimum budget requirement for Demand Gen campaigns, stating that from 1 April 2026 the API enforces a minimum daily budget of 5 USD, or the equivalent in local currency, for all Demand Gen campaigns.
The detail worth teaching is in the error handling. From API version 21 onwards an underfunded campaign returns BUDGET_BELOW_DAILY_MINIMUM; version 20 returns a generic unknown error and the real reason has to be read out of a separate field. Existing campaigns already running below the floor keep serving, but any later change to budget or duration triggers validation and has to clear the minimum first.
That last clause is the one students meet in practice. A campaign built before a rule change keeps running and looks fine, then refuses to accept an edit, and the reason is not in the interface.
Why a documented minimum is not a budget
Five dollars a day clears the gate. It does not produce a usable exercise, and Google says so in its own guidance rather than leaving it to inference. Its help page on Demand Gen campaigns recommends setting a budget at least ten times your target CPA when using target CPA bidding.
Follow that arithmetic honestly and it rules something out. If a notional target cost per acquisition is anything a real Egyptian advertiser would accept, ten times it is not a student budget. So a practice campaign cannot be a conversion-optimised campaign. Not because the student is not ready, but because the budget cannot feed the bid strategy the exercise would need.
This is the single most useful thing to understand before anyone spends anything: the budget decides which exercises are available.
What a small budget can prove, and what it cannot
| Exercise | Works on a small budget? | Why |
|---|---|---|
| Build and launch a campaign without errors | Yes | The account structure, targeting, tracking and policy checks cost nothing to get right |
| Read the diagnostics | Yes | Disapprovals, limited status and policy flags appear at any spend |
| Write and compare creative | Partly | You will see delivery differences, not statistically meaningful ones |
| Optimise toward conversions | No | The bid strategy needs far more data than the budget will buy |
| Judge a channel's cost per result | No | Any figure you get is a sample of one week on one creative |
The honest framing for a student is that a practice budget buys mechanics and judgement, not benchmarks. Anyone who finishes a course quoting a cost per lead they measured on a tiny budget has learned the wrong lesson from the right exercise.
The learning-phase story students are usually told, and what Google documents
It is common to hear that every change resets a learning period, so a small campaign never escapes it. Google's own page on campaigns limited by budget is narrower than that. It defines the limited-by-budget status as the average daily budget being lower than the recommended amount to capture the available impressions and clicks, notes the campaign can still meet its goals while missing some impressions and clicks, and states that Smart Bidding learns continuously in the background as data is collected, with budget and target adjustments taking effect immediately rather than triggering a learning status.
Teach that correction, because it changes behaviour. A student who believes every edit costs a reset learns to freeze a campaign and watch it. The documentation describes something less dramatic, and the useful discipline is changing one thing at a time for a reason you wrote down first, not changing nothing.
Who pays, and what to ask before you enrol
There are three arrangements in the market, and all of them are defensible as long as they are stated before you pay.
- The provider funds a shared account. Cheapest for you, and the campaigns are usually small and tightly supervised. Ask what you are allowed to change.
- You fund your own small budget. You keep the account and the history, which is worth more than the spend. Ask what the expected total is, in pounds, over the whole course.
- You practise on a client's or employer's account. The most realistic and the most consequential. Ask who signs off a change.
Two questions matter more than the amount. Whose billing details are on the account, and what you walk away with at the end. A campaign you built inside a shared teaching account does not travel with you, and a screenshot is not a record. If the point of the exercise is evidence you can show an employer, that evidence has to live somewhere you still control, which is the same argument as building the rest of your portfolio before you have clients.
Khaled Badr's piece on building a marketing portfolio with no experience yet takes the same problem from the other side: which projects you can run without a client at all, and what each one proves about you. Some of them need no ad spend, and a student deciding where to put a limited budget should read that before committing it.
A practice plan that does not waste money
If you control the budget, spend it in this order. The sequence is deliberate: the first two items cost nothing and remove most of the ways a small budget gets wasted.
- Build three campaigns you never launch. Structure, targeting, naming, conversion setup. Have them reviewed. Free.
- Get the measurement working first. A campaign that spends without tracking has taught you nothing but patience. Free.
- Run one campaign, one objective, long enough to be boring. Two weeks beats four separate three-day tests, because you are learning to read delivery and not to collect outcomes.
- Keep a decision log. Date, what you changed, what you expected, what happened. This is the artefact that gets you hired, and it costs nothing.
If the paid side is the reason you are taking a course at all, what a track has to cover is set out in what a performance marketing course in Cairo should teach, and the wider question of what the fee buys over free material is in free versus paid marketing courses. For judging a whole programme rather than one module, the checks are in how to judge a digital marketing diploma in Cairo.
Honest limits
Three, and the first is about this page.
We have deliberately not printed a recommended figure in pounds. We could, and it would be the most shareable line here, but a number like that depends on objective, platform, auction and month, and quoting one as though it were a rule is how students end up believing a budget is a benchmark. The documented floor above is a fact. Everything past it is a judgement that belongs to a specific exercise.
The 5 USD minimum is one campaign type on one platform. It covers Demand Gen on Google and nothing else. Other campaign types and other platforms set their own minimums, and we did not verify Meta's current figures against Meta's own documentation while writing this, so we have not stated any. Check the live page before you plan around a number you read anywhere, including here.
We sell courses, so read this adversarially. A provider writing about practice budgets has an interest in the answer. The test is the two ownership questions above. Put them to us as well as to anyone else, and if our answers are vaguer than somebody else's, that is the more useful result.