The South African CMO's survival guide to the age of Generative Engine Optimisation. And why the most sophisticated marketing play of 2026 is a very old-fashioned one.
Here’s a scenario that should keep local marketers awake at night. A prospective client types a question into ChatGPT or Google's Gemini: “Which bank offers the best customer service?”
The AI responds within seconds. It does not check your LinkedIn page. It does not read your award announcements. It does not care about the campaign you paid six figures for last quarter.
It goes to the sources it trusts: authoritative research, industry analysts, credible news coverage, Reddit and verified third-party commentary. If your brand is not being discussed there, in authoritative depth, you do not exist to this machine. And machines are increasingly making decisions on your customers' behalf.
This is not a technology story. It is a reputation story. And the uncomfortable truth I intend to argue here is this: the single highest-leverage investment a South African brand can make in 2026 is in the oldest tool in the communications kit. That tool is authoritative, human-led, editorially rigorous earned media.
Machines are quietly ranking your brand
The marketing industry has spent three years talking about generative AI as a content production tool. Cheaper copy. Faster briefs. Automated social posts. That conversation, while not without merit, has obscured a far more consequential development: the same AI infrastructure is now functioning as a brand discovery and evaluation engine for consumers worldwide.
According to the 2025 Edelman Trust Barometer Brand Report, among the 55% of global consumers who use generative AI platforms, a remarkable 91% report using them for shopping in some way. AI helps humans research brands, comparing products, and summarise reviews.
Meanwhile, Edelman's own analysis of the GEO (generative engine optimisation) landscape finds that traditional search engines are forecast to lose 50% of their search-query share by 2028 as AI interfaces absorb that traffic. The question of what the AI 'thinks' about your brand is not an academic curiosity. It is a commercial imperative.
What the AI thinks, it turns out, is shaped overwhelmingly by earned media. Research from communications agency Hard Numbers, published in their Coverage to Capital: Reputation & AI report, found that editorial media accounts for 61% of the total responses generated by LLMs when brands are evaluated – and that earned coverage was especially dominant when assessing Trust (65%) and Value (72%).
Edelman's own modelling takes this further, estimating that up to 90% of the citations that drive brand visibility in LLMs come from earned media. Your corporate blog, however beautifully search engine optimised, registers as a whisper against the authority of a well-placed feature in a respected industry title.
The big predicament for local brands
Now layer onto this global dynamic the distinctly South African consumer reality, and the stakes become acute. Our market is not experiencing a minor slowdown in brand affinity. It is experiencing a structural dissolution of loyalty.
The Truth & BrandMapp Loyalty Whitepaper for 2024-25 reveals a market paradox: 82% of economically active South African consumers – those earning above R10,000 per month – are enrolled in at least one loyalty programme, up from 76% the year prior. That sounds like an engagement success story. It is not. Those same consumers hold an average of 10.3 active loyalty programme memberships, cherry-picking benefits with no emotional attachment to any single brand. Loyalty scheme participation has become an act of arbitrage rather than allegiance.
At the lower end of the income spectrum, the picture is bleaker still. The 2025 Township CX Report, produced by Rogerwilco in partnership with Field & Insights Africa, found that 39% of township shoppers switch brands the moment prices rise. This figure doesn’t reflect fickleness but survival. “Conventional marketing wisdom assumes loyalty is fixed,” observes senior brand strategist at Rogerwilco, Mongezi Mtati. “What our research shows is that loyalty in the township context is highly rational and deeply tied to survival.”
What this means for brand leaders is that the traditional retention playbook, more points, more app notifications, more personalised offers, is becoming structurally inadequate. You cannot buy your way into a consumer’s mental shortlist when the mental shortlist is being curated by an AI that has never seen your promotional emails.
The path to durable consideration in this market runs through credible, authoritative third-party verification – the kind that earns LLM citations and shapes latent perceptions of AI long before a consumer types their first question.
The misinformation tax no one budgets for
There is a second, under-appreciated threat compounding this challenge: the corrosive effect of AI-generated misinformation on brand credibility. The World Economic Forum’s Global Risks Report 2025, drawing on the views of over 900 global experts and policymakers, ranked misinformation and disinformation as the number one short-term global risk for the second consecutive year.
The WEF is explicit that AI tools are enabling a proliferation of false information in the form of video, images, voice, and text, and that the resulting erosion of epistemic trust is 'becoming more difficult' to counter with each passing month.
The Reuters Institute’s Digital News Report 2025, based on nearly 100,000 respondents across 48 countries, found that 58% of global news consumers are worried about what is real and what is fake online. Critically, that figure rises to 73% in Africa, the highest among the regions surveyed, where social media saturation and variable media literacy create fertile ground for synthetic content to take root.
This creates a dynamic I think of as the Misinformation Tax: a mounting premium that brands must pay in credibility-building activity to maintain baseline trust in an environment awash with synthetic content. Brands that are not proactively seeding accurate, authoritative information into the media ecosystem – through journalists, analysts, and credible trade outlets – are not standing still. They are falling behind, ceding ground to whatever version of their reputation the algorithm constructs in their absence.
The machine-friendly marketer's toolkit
None of what I have written above is an argument against AI in marketing. It is an argument against AI as a substitute for strategic communications. The brands that will win this decade are those that use AI to augment and distribute human insight – not to generate the appearance of insight at scale.
The Edelman Trust Barometer provides a clear mandate for what human insight must deliver. Across 15 markets and 15,000 respondents, the 2025 report found that 80% of consumers trust the brands they use more than they trust governments, the media, or NGOs. That is a remarkable degree of power – and a remarkable degree of responsibility. But it comes with a condition: 73% of those consumers say their trust would increase if a brand authentically reflected today’s culture.
In a South African context, that is not a request for a Zulu-inflected social media campaign. It is a demand that brands demonstrate a genuine, evidence-based understanding of the economic and social realities their consumers are navigating, and that understanding be validated by sources the consumer already trusts.
In practical terms, this translates to a reorientation of the communications function that BOLD has been advocating for years. Stop commissioning content. Start commissioning journalism. The distinction is not semantic. Content serves a brand’s ego. Journalism serves an audience’s need – and in doing so, earns the credibility signals that AI will recognise and amplify.
Our own analysis found a 30-fold differential in readership between generic PR outputs and pieces that delivered locally grounded, contrarian, evidence-led insight. An awards announcement commands 36 reads. A rigorous analysis of the South African brand loyalty crisis commands over 1,000. The LLMs are not the only ones making this distinction; your audience already has been.
Four editorial principles separate brands that are building AI authority from those that are flooding the market with forgettable output:
- Lead with the uncomfortable truth. The highest-performing thought leadership identifies a threat the audience has not yet fully articulated and frames it within their specific commercial context. Not 'AI is changing marketing'. Rather: 'Your brand does not exist to the AI making decisions for your customers, and here is the evidence.' Provocation is not rudeness. It is editorial precision.
- Anchor every claim in verifiable, named research. The Reuters Institute. The Edelman Trust Barometer. The WEF Global Risks Report. These are not decorative citations. They are authority signals that LLMs are trained to weigh. A thought-leadership piece supported by these sources is not just more persuasive to a human reader; it is more likely to be indexed as credible source material in the AI’s knowledge graph.
- Localise without patronising. The global trend is the entry point, not the destination. Every piece of South African brand communications should close the gap between macro research and the specific lived experience of a market characterised by 39% brand switching due to price sensitivity and loyalty programme fatigue among the professional class. Generic GEO advice is table stakes. What a Johannesburg retailer, an Afrikaans financial services brand, or a Cape Town tech firm needs is the local application. That is where the value and the differentiation live.
- Turn client wins into industry lessons, not trophy cases. Nobody reads a case study that begins with 'we delivered exceptional results'. Everyone reads a case study that begins with 'our client was losing 15% of its customer base annually, and the conventional wisdom about how to fix that was wrong'. The former serves your sales deck. The latter serves the industry – and in doing so, earns the kind of third-party discussion and amplification that the AI will find, index, and cite.
The bottom line
South African brand leaders are operating in one of the most demanding communications environments on earth: a consumer base under sustained financial pressure, a media landscape fragmenting faster than budgets can keep up, and an AI infrastructure silently rendering verdicts on brand credibility in real time. The temptation to automate – to generate more, spend less, and 'fill the funnel' with AI-produced content – is understandable and, in the short term, seductive.
It is also a category error. The AI is not impressed by volume. It is impressed by authority. And authority, in 2026, is still built the way it always was: through human expertise, independently verified, published in outlets that have earned the trust of an audience and, by extension, the machine's trust.
The gap between a brand that gets 36 reads and one that gets 1,000 is not a budget gap. It is an editorial courage gap. In the age of generative AI, that courage has a new, measurable consequence: the difference between a brand the AI knows and one it ignores.
The robots are watching. Make sure they find something worth citing.
