The best evidence for personalization is not a survey about what customers say they want. It is the gap between what a triggered message earns and what a broadcast earns from the same list. Klaviyo’s benchmark study of more than 143,000 abandoned cart flows found an average of $3.65 in revenue per recipient, with the top decile at $28.89. No merge tag produces that. Timing and context do.
The name is not the personalization
A first name in a subject line is a formatting choice. It carries no information the recipient did not already have, it is trivially copied by every competitor and every phishing attempt, and when the merge fails it prints "Hi {FirstName}" and tells the reader exactly how automated the relationship is. The reason it persists is that it is the only kind of personalization that requires no thought about what to send.
What actually changes behaviour is a message whose existence is justified by something the recipient did. The abandoned cart email works not because it says "Hi Sarah" but because it arrives while the intention is still live and references the specific thing that was left behind. Strip the name out entirely and the performance barely moves. Strip the context out and it collapses.
A ladder you can actually climb
| Level | What it needs | Realistic effect |
|---|---|---|
| Merge fields | A name field, populated and validated | Negligible, and negative when the merge breaks |
| Lifecycle segmentation | Signup date, purchase count | Meaningful. A first-week subscriber and a three-year customer should not get the same email |
| Behavioural triggers | Event tracking from your site or store | Large. This is where the revenue-per-recipient gap lives |
| Dynamic content blocks | A rules engine plus enough catalogue or content to vary | Moderate, and it compounds with the level above rather than replacing it |
| Predictive scoring | Volume most small lists do not have | Unproven at small scale. Skip until the levels above are exhausted |
Effect sizes are directional. The abandoned cart figures cited come from Klaviyo’s 2024 benchmark report covering flows sent in 2023.
Note the order. Most businesses attempt dynamic product recommendation blocks before they have built a single behavioural trigger, because recommendations feel more sophisticated. They are not. A recommendation block inside a broadcast that goes to everyone on a Tuesday is a broadcast with variable images. A plain-text email triggered by a specific action is personalization.
Use the data you already have, which is more than you think
Before buying anything, inventory what your existing tools already record: signup source and date, which form or landing page converted them, purchase history and order value, category or product pages viewed, which past emails were clicked, and support tickets raised. Signup source alone is underused and powerful. Someone who joined from a pricing page and someone who joined from a blog post about a beginner topic want visibly different things.
One caveat on the engagement data specifically: build your rules on clicks, site visits, and purchases rather than opens. Apple Mail Privacy Protection prefetches tracking pixels for a large share of recipients, so "opened but did not click" is not a meaningful state to personalize against. See How to Measure Email Marketing ROI for the detail.
Lawful basis is part of the design, not a footnote
If you are profiling EU subscribers to decide what to send them, GDPR Article 6 requires a lawful basis for that processing, and whichever basis you choose has to be documented and disclosed in your privacy notice before you rely on it. If you rely on consent, Article 7 requires that you be able to demonstrate it and that withdrawal is as easy as giving it. This is not merely a legal chore: a preference centre that lets someone opt out of behavioural targeting while staying subscribed is both the compliance mechanism and the retention mechanism.
Where personalization stops paying
Every conditional block you add is a template variant that has to be QA tested, and the number of paths grows multiplicatively. A four-condition email has sixteen possible renderings, and one of them contains an empty div that ships to a customer. The practical ceiling for a small team is a handful of triggers, each with a small number of variants, all of which someone actually previews before send. Beyond that point the maintenance cost eats the lift, and unlike the lift, the maintenance cost is certain.
Key takeaways
- Behavioural triggers, not merge fields, produce the revenue gap. Klaviyo measured $3.65 average revenue per recipient on abandoned cart flows.
- Climb the ladder in order: lifecycle segmentation and triggers before dynamic content blocks and predictive scoring.
- Signup source is the most underused field you already have, and it predicts intent better than a name field.
- Build rules on clicks and purchases, not opens, which Apple Mail prefetching has made unreliable.
- Every conditional block multiplies the renderings you must QA. The ceiling is set by what you can actually preview.
Related reading
Sources
- Abandoned Cart Benchmark Report, Klaviyo (2024)
- Article 6 GDPR: Lawfulness of processing, GDPR text
- Article 7 GDPR: Conditions for consent, GDPR text
- Email Client Market Share, Litmus
- Personalization
- Email Marketing
- Conversion
Written by
Valter Brandt
Chief Marketing Officer
Valter Brandt is the Chief Marketing Officer of ThisCom, working with clients across the United States and internationally. He has led marketing strategy through the major shifts in social advertising, mobile, content marketing, programmatic media, and marketing automation.
All articles by Valter Brandt →
