- Digital campaign metrics often misrepresent real business impact
- Ad fraud is widespread, scaled and underreported
- Verification tools lag behind evolving fraud tactics
- Common-sense analytics can outperform complex tools
- Forensic auditing can recover 20%-60% of wasted ad spend
For marketing professionals running digital campaigns, the following scenario is likely familiar: The campaign report arrives and your metrics, whether landing page traffic, click-through rates, or leads generated, look good vs your benchmarks.
And the fraud detection tool gives you a clean bill of health, with invalid traffic at a reassuringly low level and ‘viewability’ where you’d want it to be.
Business outcomes not aligning with campaign results
Then comes the sales team meeting. Despite brilliant digital ad results and record digital ad spending, footfall is flat. Online conversions have not increased and the product has not shifted. Something, somewhere, does not add up.
This scenario plays out in marketing departments across South Africa every single week. It is not a symptom of naivety or incompetence. It is a structural problem baked into the way digital advertising is bought, measured and reported.
Research shows that only about one in every four online ad impressions is actually viewed by a real person. Globally, digital ad fraud losses are projected to reach $172 billion by 2028, an incomprehensible number.
That is not a niche problem for multinationals, and it’s happening to all online advertisers, whether extremely small or the biggest spenders in every country. It affects every brand spending, especially on programmatic inventory, from a Sandton financial services giant to a Cape Town e-commerce start-up.
The uncomfortable truth is that your dashboard is probably showing you what it was designed to. And this is not necessarily the absolute, forensic reality.
Ask the obvious questions
The rise of programmatic advertising brought something genuinely useful: scale and efficiency. It also brought something less welcome: complexity so dense that measurement systems failed to keep up with the unintended consequences of measurement, leaving marketers with only part of the picture.
Dr Augustine Fou, a cybersecurity researcher and the founder of FouAnalytics, has spent nearly three decades studying exactly this problem. His conclusion is disarmingly straightforward. “Solving ad fraud,” he says, “is not as hard as most people think. If they actually wanted to solve it, they would basically pay attention to their own analytics. Look for tell-tale signs of fraud. These things just don’t make common sense.”
The key phrase here is common sense. Not machine learning. Not a seven-figure verification contract. And it doesn’t need an analytics suite.
Fou is equally blunt about the limits of the tools many marketers rely on. Legacy verification systems are typically built to detect the fraud that was common when they were first designed. Sophisticated bad actors have long since evolved past those defences, while the fraud rate reported to clients stays suspiciously low and suspiciously stable, year after year.
The question worth asking is not, “Does my tool say I have been defrauded?” It is, “Do these numbers make sense for a real human audience?”
The common-sense ad fraud checklist
You do not need a PhD from MIT to start applying forensic thinking to your campaigns. You need four things: an eye for behaviour, a feel for numbers, a basic knowledge of your inventory, and the willingness to exercise your own judgement.
1. Behaviour: Does this look like a human being?
Real people are lazy, distracted and impatient. They do not, in high numbers, click on ads while playing mobile games; they do not visit obscure recipe websites at three in the morning; they do not bounce off a landing page in 0.2 seconds and then return to click again four seconds later; they do not stare at the screen of audio players; and they do not stare at their alarm clock apps all night.
Fou illustrates this elegantly. Think about someone playing Candy Crush on their phone. An ad appears. What is the likelihood they click through, visit an e-commerce site, and complete a purchase while mid-game?
“It just doesn’t happen in real life,” he says. Yet click-through rates on certain mobile game inventories can look spectacular on paper. That is because the clicks are often not human choices. They are the result of accidental taps, hijacked gestures, or outright bot behaviour.
Practical test: Check your time-on-site data for traffic arriving from your programmatic buy. If the average session duration is mere seconds, something is wrong. Real South African consumers browsing a retail or financial services site stay longer than that, even when they are uninterested.
2. Numbers: Does this volume make sense?
Fraud operates on the logic of arbitrage. Bad actors buy cheap bot traffic for a website, then sell the ad impressions on that website at a higher CPM. The math works because the scale is enormous. And enormous scale is exactly what should raise your eyebrows.
“There’s not a whole bunch of humans sitting around with nothing to do but to go to your website when you tell them to,” Fou notes. Yet advertisers have become, in his words, “addicted to irrational quantities.” When a niche B2B brand serving financial advisers is offered fifteen million impressions per month at a very attractive CPM, the right response is not excitement. It is scepticism.
Practical test: Divide your impressions by your total addressable audience. If the ratio means the average person in your target segment saw your ad fourteen times in a week, ask yourself whether that is plausible or whether some of those impressions were going to people and bots who were never your customers.
3. Inventory: Do you know where your ads are running?
Fraud hides in plain sight, dressed up as legitimate inventory. Domain spoofing, where a fake site masquerades as a premium publisher, is widespread precisely because it is effective. A bid request that claims to be a respected news site will attract advertiser bids. A bid request that reveals a randomly generated domain will not.
Fou points out that the “top mobile apps by ad revenue are entirely different from the ones humans use most.” That gap between the apps humans love and the apps raking in ad spend is, very often, the gap where fraud lives.
A simple example: A local retail brand running programmatic display may find a portion of its budget landing on templated content farms with names that sound plausible but attract no organic traffic whatsoever. Pulling a simple placement report and Googling the top twenty domains takes twenty minutes. It is twenty minutes well spent.
A practical test: Request a full, and this is very important, post-bid placement report from your buying platform. Review the top fifty domains by impression volume. Visit them. Read them. Ask whether a real South African customer would ever find themselves on that site organically.
4. Judgement: Would you believe this if it were your own money?
This is the most important test of all, and the one most consistently skipped. Marketers reviewing agency or vendor reports often apply a professional generosity they would never extend to their own household finances.
If a friend told you they had invested their savings into a scheme that was generating a 40% monthly return with zero risk, you would ask uncomfortable questions. Apply the same instinct to a campaign report showing 98.8% valid traffic, a 6% click-through rate, and minimal conversions.
Practical test: Before signing off on a campaign report, ask one question aloud: “If this were my own money, would these numbers convince me?” If the answer requires any kind of rationalisation, that is your signal to dig deeper.
5. Website (landing page) analysis
(Using GA/Adobe Analytics will unfortunately limit traffic type analysis, as both block bots.)
Traffic patterns: Look for anomalous spikes in traffic and activity, such as abandoned baskets, form fills, account creation, or higher-than-expected volumes, particularly in the hours when people are sleeping. See, if possible, what channels these correlate to.
Bounce rates: These relate strongly to bot behaviour, not merely mismatches in user expectations and website content. What patterns do you see?
When advertisers start asking sensible questions about where their money is going, the winners are the platforms with real audiences, auditable traffic, and transparent reporting. The losers are the anonymous inventory mills that inflate numbers and deliver nothing.
Local publishers and broadcasters have a genuine opportunity here. Offering transparent, log-level reporting to clients, proactively sharing suspicious domain lists across the industry, and educating brand clients about what quality inventory actually looks like are commercial advantages.
In a market where buyers are increasingly scrutinising quality, being the publisher who welcomes scrutiny is a powerful differentiator.
No good platform is devoid of bots that inadvertently cause ads to land. And then there are the ‘good bots’ that help measure and analyse traffic. FouAnalytics has a solution for this: a tag that analyses ‘bot vs not’ and if it ‘is bot,’ then an ad call is not made
The common-sense campaign audit
You do not need to overhaul your tech stack. You do not need a six-month procurement process. You need one campaign, one placement report, and the willingness to ask a few straightforward (but perhaps awkward) questions.
Pick a campaign that ran in the last quarter. Pull the placement data. Apply the four tests above: behaviour, numbers, inventory, and judgement. Note what raises a flag and what does not. Then compare those findings with what your verification tools told you at the time.
The gap between the two pictures is your opportunity to recover your budget. For most South African advertisers, it is a meaningful one. Forensic media analytics consistently shows that brands recover between 20% and 60% of their media spend once they shift from inflated, low-quality inventory toward verified human audiences.
That is not a loss to grieve. It is budget to redirect. Funds that have been quietly disappearing into fake traffic can instead be pointed at the real people who actually buy things.
Dr Fou puts it simply: “They’ll actually be able to solve a lot of the fraud themselves. What I’m trying to do is educate them on how to do that so that they can be part of the solution.”
If you got 20% to 60% of your online advertising budget back, how would you invest it?
Marc Dhalluin is the co-founder of TruthsetsOnline.com,.With over 25 years in marketing — spanning agency, client, and consultancy roles across Africa, Europe, and the US — he has helped build, transform, and rescue brands, generating more than $500 million in annual revenues. A long-time collaborator of fraud researcher Dr Augustine Fou, Marc is now focused on one thing: proving that real advertising reaches real humans. He is based in Los Angeles and writes regularly on digital transparency, programmatic fraud, and what brands should actually be measuring. Connect with Marc on LinkedIn or visit TruthsetsOnline.com
