The controversial Cambridge Analytica case, which revealed the use of psychographic data to manipulate voter behaviour in the 2016 US elections, sparked Hove’s interest in the world of online political advertising. The scandal rang alarm bells, but Hove remained sceptical: How effective are these practices really? Are campaigns as strategic and data driven as media narratives suggest?

Millions of political advertisements

To find out, Hove and colleagues analysed millions of political advertisements on platforms like Facebook and Instagram, across close to 100 countries. He investigated how political campaigns use these targeted ads to communicate with voters and what their effects might be.

‘Many of these ads are so-called “native advertisements” – they blend in with regular content and are only subtly labelled as “sponsored”,’ Hove explains. ‘Most users likely don’t notice these labels, especially when scrolling quickly through platforms like TikTok, Instagram or Facebook. That makes it harder to critically assess political content.’

Not so ‘micro’ after all

Hove found that while microtargeting is certainly used, it is far less sophisticated than often assumed. ‘We talk about “microtargeting”, but the targeting is not so micro. Most political ads are targeted using basic demographic criteria – most commonly age and geographic location. The strategies are relatively simple and lack the psychographic complexity once associated with Cambridge Analytica.’

There’s also little evidence that these ads are effective in persuading people to switch parties. ‘At least not in terms of converting someone from left to right or vice versa,’ says Hove. ‘Parties mostly focus on energising their existing base. So as a voter, you're likely seeing messages that reinforce what you already believe – and little from parties you might not (yet) support.’

One country, many campaigns

Still, the impact of microtargeting is real. ‘Messages are subtly tailored to fit specific groups. For example, younger voters are more likely to see ads about education, while older voters see more about healthcare or inflation – regardless of whether that topic fits the party’s traditional profile. And in terms of geography, urban areas tend to get more centre-left ads, while rural areas see more from right-wing parties.’

Hove warns that such selective messaging can deepen social divides. ‘If people consistently see only the issues that apply to their own demographic, they may begin to think that political parties only care about those issues. That can create perception gaps between groups and amplify existing tensions – not because policies actually differ, but because messaging is selectively delivered.’

The AI wildcard

Hove also examined how generative AI could amplify the power of microtargeting. ‘Most people struggle to tell AI-generated images apart from real ones. That opens the possibility for campaigns to deploy fabricated images or videos to reinforce claims or discredit opponents.’

As these tools become more accessible it may become harder and harder for voters to tell what’s real. ‘The future of political communication may look very different – and potentially be more dangerous – than today’, he warns.

Voters should stay vigilant

Hove concludes with a call for media literacy and critical thinking. Not all political advertising on social media is harmful, he says – it can help elevate marginalised voices. But voters should stay vigilant and seek out trustworthy journalism, especially during elections. ‘Subscribing to credible news outlets and reading full-length articles – rather than just headlines or memes – is essential for forming well-rounded political opinions.’

Thesis details

Mads Hove, 2025, The Nature and Consequences of Online Political Microtargeting. Supervisors: Prof. C.H. de Vreese and Prof. S.B. Hobolt. Co-supervisor is Prof. A. van Dalen.

Time and location

Friday July 11, 2025, Agnietenkapel, Universiteit van Amsterdam