Every marketing team eventually asks some version of the same question: how much should go to search, how much to social, how much to everything else? It is a reasonable question and it has no useful answer in the abstract. The splits that get quoted in articles are averages of businesses that have nothing in common with yours.
What does transfer is the order of decisions. Get the sequence right and the percentages resolve themselves. Get it wrong and you will spend the year defending a number you picked before you understood the problem.
Why the imported rules travel badly
The 70/20/10 convention, and the various brand-versus-performance ratios that circulate alongside it, were developed in markets with different media economics, different mobile data costs, different retail geography and different language segmentation. South Africa breaks several of the assumptions underneath them.
- Data cost still shapes behaviour. Video-heavy strategies behave differently in a market where a meaningful share of the audience is managing a data budget, and that changes creative length and placement choices before it changes the split.
- The audience is genuinely multilingual, and language is a casting and media decision rather than a translation task at the end.
- Commuting and retail geography make out of home materially more valuable here than the digital-only playbooks assume.
- Category concentration is unusual. A handful of large advertisers dominate certain categories, which distorts auction pricing in ways an international benchmark will not predict.
The four questions, in order
1. What is the commercial objective, precisely?
Not the marketing objective. The commercial one. Acquiring first-time buyers, defending share against a specific competitor, launching into a category that does not know it exists yet and shifting stock before a season ends are four completely different allocation problems that all get described as awareness.
The test of a real objective is that it tells you what to stop doing. If your stated objective does not rule anything out, it is a mission statement.
2. Where is the audience, actually?
Not where it is convenient to buy them. This is the point at which most plans quietly narrow to the two self-serve platforms the team already has logins for, and the plan starts describing the agency's comfort rather than the customer's day.
Audience definition should be specific enough to be wrong. Region, language, commute, income band, what they do at eleven on a Tuesday. If the definition would fit half the country, it will produce a media plan that reaches half the country expensively and none of it persuasively.
3. What does the brand already own?
Owned audience, an email list, a site that ranks, a community, a set of creators who already talk about you unpaid. These are budget the brand does not have to spend, and they are systematically undervalued because nobody invoices for them.
This is also where the largest available saving usually sits. Paying to reach an audience you already have access to is one of the most common and least noticed forms of media waste.
4. What can you actually measure?
Be honest about attribution before you allocate against it, not afterwards. If a channel's contribution cannot be measured with the tracking that genuinely exists, it should be funded as an investment with a defined review point rather than defended weekly with numbers nobody believes.
Split the budget, then split it again
The most common structural mistake is treating the media budget as one number. It is at least three: working media that buys audience, production that makes what runs in it, and the fee for planning, negotiating and reporting. Brands that do not separate these cannot tell whether a disappointing quarter was a media problem, a creative problem or an agency problem.
Creator campaigns need the same treatment. Creator fees, production and paid amplification are three lines, and campaigns that allocate nothing to amplification consistently underperform, because the content stops working the moment its organic reach does.
Then leave it alone long enough to learn something
The opposite failure to setting the split in stone is changing it every fortnight. Brand effects are slow and performance effects are fast, so a plan judged only on the fast signal will drift towards the bottom of the funnel until there is nothing left feeding it.
Agree upfront how long each part of the plan gets before it is judged, and what evidence would justify moving money early. Written down, before the first week of results makes everyone brave.