AI-Personalized vs. Generic Emails: UK School Decision-Maker Response Experiment | EducationDataLists

These numbers should not be interpreted as guaranteed benchmarks for cold B2B outreach to school decision-makers. They primarily provide a picture of the broader UK email environment.

The important lesson is that delivery and attention are different problems. A technically deliverable email can still fail to create a conversation if the subject, offer, timing, or message is irrelevant.

MailerLite's 2025 benchmark analysis similarly reported a 1.93% median click rate for educational institutions, compared with 2.09% across all industries. (MailerLite)

For marketers, this suggests that campaign success should not be evaluated using open rates alone. Clicks, replies, qualified conversations, meetings, and downstream opportunities are more meaningful measures of commercial or partnership outreach.

AI-Personalized vs. Generic: What Does the Evidence Suggest?

A genuine controlled experiment would require two comparable segments of UK school contacts receiving different versions of the same campaign. Without such proprietary test data, it would be misleading to claim a specific percentage lift in replies for AI-personalized messages.

However, broader email research provides strong evidence that personalization can materially affect engagement.

MoEngage's 2025 benchmark study analyzed more than 17.3 billion marketing emails and compared attribute-, journey-, and behavior-based personalization with broadcast emails. Its analysis found that personalization based on behavior could produce substantially higher engagement and conversion outcomes than non-personalized campaigns. (MoEngage)

The exact magnitude varies significantly by industry, customer journey, and personalization method, so these figures should not be transferred directly to UK school outreach.

The more defensible conclusion is that personalization works best when it reflects actual recipient context.

A Practical Experimental Framework

A marketer testing a UK Schools Email Database could structure an experiment like this: