
Personalisation has looked like the holy grail of digital marketing for years. The logic is seductive: the better you know someone, the sharper your message, the bigger the effect. And online you can know just about everything, right?
The research began with a promise
In 2017, Matz and her colleagues published three field experiments with psychologically tailored ads in PNAS. Reach? More than 3.5 million people. Those who received a message that fitted their psychological traits clicked and bought significantly differently, and the authors duly noted it, significance tests and all.
The research was more than a lab exercise. Everything happened on the internet and the behaviour was real, not just a filled-in ten-point scale. Only... co-author David Stillwell made his money with a personality test on Facebook, and although the authors report that themselves, it is a bit like the Belgian railways researching the benefits of public transport.
What the meta-analyses say
Of course there are more studies hunting for the how, the what and the why of that holy grail. A 2022 meta-analysis looks for the common thread. Joyal-Desmarais and colleagues put 702 experiments side by side, with more than 206,000 participants in total.
Those who tailored their message to what people want to achieve saw, on average, a clear but modest effect. Other forms of tailoring (so addressing people on personality traits) often did worse. You get the most out of tailoring to the situation.
How do you know whether you are really addressing people personally? Online, everything is readable. Cookies, tracking and other first- or third-party data promise it. Perla and colleagues pooled 41 studies from marketing, psychology and computer science (Psychology & Marketing, 2026) to find out how well those predictions can be made.
From digital traces, about 5 percent of the variance in personality can be predicted, they note. Five percent. Read that again. The horoscope in a weekly magazine in the doctor's waiting room does better. Keep in mind that personalised messages have a measurable but small effect on behaviour, then take into account the weak spots in the design of some (lab) studies, and the total effect of psychological targeting comes down to practically zero.
The opposite effect
The Attitudes and Persuasion Lab at Ohio State pooled decades of research on personalisation and concluded that people drop off when ads feel like an invasion of their privacy, like an attempt to steer them, or like a stereotype stuck on them. A message they already knew no longer touches them. And a weak argument does not get stronger through tailoring, it simply stands out more.
What does work
Personalisation works when you tailor the message to the situation and to the motivation of the people you want to reach. Anyone who tries to guess a personality from a trail of digital pebbles is building on fortune-telling and horoscopes.
For your next briefing, then, there are three questions. Which decision is this person facing right now? What do they want to achieve? What do they already know? Ask these first. That is where you find the grail of personalisation.