Personalisation adapts content to a specific person using data you hold about them. That is the line that separates it from dynamic content, which changes by context and needs nothing personal at all.
Where it pays is in video. A personalised video that addresses the customer by name, reflects their actual plan or basket, and answers the question their account raises is a fundamentally different object from a generic explainer, and it can be generated per customer rather than filmed per customer.
The caveat is the gap between relevant and unsettling, and it moves with context. Referencing something the customer told you directly reads as attentive; referencing something you inferred reads as surveillance, even when both come from data you hold lawfully. The reliable test is whether the customer would be comfortable being told how you knew.
Second, personalisation multiplies quality problems. One video with an awkward sentence is a mistake; fifty thousand generated from the same template is a campaign-wide one. Review the template as if every variant will be seen, because collectively they will be.
The blog covers the practice side: balancing scale and authenticity.