- South African ad fraud: Fraudlogix’s 2026 report found that 22.94% of South African ad impressions analysed from 2025 showed characteristics of non-human or fraudulent activity, above the global rate of 20.64%.
- Bots outnumber human web traffic: Imperva’s 2025 Bad Bot Report found that automated traffic accounted for 51% of global web traffic in 2024, with malicious bots representing 37%.
- Ad impressions do not guarantee human attention: An ad being served or its pixels loading does not prove that a real person saw it, raising questions about the reliability of digital advertising reach and viewability metrics.
- Ad fraud affects marketing returns: TruthsetsOnline, powered by FouAnalytics, says its forensic analysis of more than 100 South African campaigns covering close to one billion impressions routinely detects fraud rates of 20% to 60%, depending on campaign setup.
- Independent verification can improve accountability: TruthsetsOnline co-founder Marc Dhalluin argues that advertisers should independently verify media delivery to identify fraudulent impressions, improve governance and redirect wasted advertising spend towards real audiences
As the region’s creative industry celebrates at the Loerie Awards in Cape Town, forensic media analytics company TruthsetsOnline raises a glass and asks a question no category judges can answer: outside the jury room, how many real human beings saw the work?
Independent research suggests that more than one in five ad impressions in South Africa shows signs of non-human activity.
The Loerie Awards, held in Cape Town on 8 and 9 October 2026 as the finale of Loeries Creative Week, have once again demonstrated that agencies in Africa and the Middle East can create advertising that stops people in their tracks.
Every winning entry had one thing in common: a jury of real people who sat down and watched it properly. TruthsetsOnline, the forensic media analytics company powered by FouAnalytics, wants to ask the awkward follow-up: Beyond the jury, how many real people saw it?
Nobody can say for sure. And co-founder Marc Dhalluin says that’s the point.
“The creative work at the Loeries is world-class, and nothing we say detracts from it,” says Dhalluin. “Our question is about the plumbing, not the paint. A brilliant ad shown to a bot – let alone hidden – is still brilliant. It’s just a very expensive one.”
Birdwatchers would never accept this evidence
The loerie is a bird (be it the colourful Tauraco corythaix or the grey Crinifer concolor), so here is a birding comparison. A serious birder will not log a sighting on someone else’s say-so. If they did not see it, it does not go into the field diary.
Digital advertising records its sightings on far weaker evidence. It records an impression when an ad is served, and a view when pixels load. Neither confirms that a person was on the other side of the screen.
Platforms paid to deliver ads usually provide that confirmation. Legacy tools used by agencies and brand teams typically underreport scope and scale of fraud by orders of magnitude.
The research suggests the gap is large
- More bots than people. Imperva’s 2025 Bad Bot Report found that automated traffic overtook human traffic on the web in 2024, accounting for 51% of all traffic, with malicious bots making up 37%.
- South Africa is above the global average. Fraudlogix’s State of Ad Fraud 2026reportanalysed 105.7 billion ad impressions from 2025 and found that 22.94% of South Africanimpressions exhibited characteristics of non-human or fraudulent activity, compared with 20.64%
- The filters miss a lot. In 2024, independent researchers at Adalytics reported that a widely used bot-filtration product classified declared bots as valid human traffic 17% of the time and undeclared bots 77% of the time. Note that the vendors named disputed the findings.
- Governance changes the outcome. The ANA’s Q4 2025 Programmatic Transparency Benchmark found that advertisers with strong quality controls turned 56.7% of programmatic spend into fraud-free, viewable, and measurable impressions, compared with 37.5% for those without strong controls.
- South African campaigns analysed. We have forensically analysed more than 100 South African campaigns, close to 1 billion impressions, and the numbers hold fast. We routinely detect 20-60% fraud using Fou Analytics (campaign setup dependent).
There is no category for the audience that wasn’t there
The 2026 Loeries category list runs from Digital and Media Innovation to Effective Creativity, with a new Creative Strategy category this year. Judges assess results-based entries against the evidence presented, which often relies on reach, views, and impressions. None of those numbers alone prove a human was present.
“If there were a Loerie for the most convincing audience that doesn’t exist, the bots would sweep the board,” says Dhalluin. “They never skip an ad, never complain about frequency, and are cheaper to reach than any human. Which is exactly why a campaign optimised for high volume and cheap reach keeps finding them.”
Why it matters now
For marketers and brands, the case is commercial: money spent reaching bots is money that did not reach customers.
For boards, the question is becoming a governance issue. From 15 December 2026, ISA 240 (Revised) sharpens the expectation that auditors apply professional scepticism to fraud risk, and digital media is often one of the largest discretionary lines in a listed company’s accounts.
“Nobody needs to feel unworthy of a trophy,” says Dhalluin. “The creative did its job. The question is whether the media did its job too, and we can check that. Marketers who review their own numbers usually find money to reinvest in front of real people.”
TruthsetsOnline offers marketers a free forensic audit, run alongside their existing measurement tools, so they can compare results impression by impression. Marketers who would like to see their own numbers can contact Marc Dhalluin via InMail or email.
