Advertising · Randomized field experiments
Personalization in Email Marketing: The Role of Noninformative Advertising Content
Marketing Science · 2018
Summary
In the main experiment, adding the recipient’s name to the subject line raised the open probability from 9.05% to 10.80%, increased sales leads from 0.39% to 0.51%, and reduced unsubscribes from 1.2% to 1.0%.
For your marketing
Possible founder application
A strong example of testing a narrowly defined personalization change with downstream outcomes. For a current campaign, randomize the treatment and measure outcomes that matter beyond opens.
Keep the offer, audience and email body the same while testing the subject-line change. Choose a downstream outcome such as qualified leads or purchases, and track unsubscribes as a guardrail. Use opens as supporting context rather than the only success measure; a subject line that attracts attention still needs to produce useful business outcomes.
Reported result
What the study found
In the main experiment, adding the recipient’s name to the subject line raised the open probability from 9.05% to 10.80%, increased sales leads from 0.39% to 0.51%, and reduced unsubscribes from 1.2% to 1.0%. Similar experiments at other firms found benefits across acquisition and retention objectives.
Research question
What the authors investigated
Does adding consumer-specific but noninformative content to an email affect recipient response?
Design
Randomized field experiments with three companies selling different products; email messages were tailored with details such as recipients’ names.
Sample and setting
The experiments sent messages to millions of individuals. The abstract reports detailed results for a main experiment and replication across different advertiser objectives; it does not give a single pooled sample size.
Scope and transfer
Limitations to keep in view
Effects are from specific companies, messages, audiences and campaign conditions. The main treatment is name insertion, not deep behavioral personalization. Email measurement and privacy practices have changed since these campaigns; the percentages should not be used as current benchmarks.
Research sources
Publisher and Stanford GSB abstract; detailed full text was not available in the retrieved pages. The summary is restricted to the reported abstract results.