Knowledge base
Cross-channel personalization at scale (and where AI fits)
Personalization at scale means tailoring content, timing, and channel for each person automatically, without hand-writing a million messages. It runs on two different engines: templating and AI. Keeping them straight is the whole skill.
7 min read · Updated June 28, 2026
Dynamic content: the templating engine
The first engine is plain templating. A template carries placeholders that get filled per recipient at send time: a first name, the items left in a cart, the points balance, a recommended product from a catalog. Iterable uses Handlebars syntax for this, which also supports conditionals, so one template can render very different messages for different people. No AI is involved; it is structured substitution.
This covers most of what people call personalization. It scales because the logic lives in one template, not in a thousand hand-edited emails.
The AI layer: predict versus generate
The second engine is AI, and it splits cleanly into two kinds that are easy to confuse.
- Predictive AI forecasts a number or a choice from past behavior. In Iterable this includes Send Time Optimization (the hour each person is most likely to open), Channel Optimization (the channel they engage with most), Frequency Optimization (how often before they tune out), Brand Affinity (a sentiment label per user), and Predictive Goals (how likely each user is to hit a goal you define).
- Generative AI writes new content. In Iterable this includes Copy Assist (suggests subject lines and body copy) and journey-building assistants that draft cross-channel flows from a prompt.
The line to keep crisp
Predictive AI decides when, which channel, and to whom. Generative AI decides what to say. Calling send-time or channel optimization "generative AI" is a credibility error: those are predictive models, not text generators.
Saying it accurately
Vendor AI branding moves fast. Iterable has recently grouped its AI features under a "Nova" suite on its marketing site, while its support documentation still uses the classic names above. Both describe largely the same optimization stack, so the honest move is to use the documented capability names and confirm the current generally-available product names with the team rather than reciting a marketing page.
The SE instinct
Describing a capability accurately and then saying you would verify the current GA naming is not a weakness in a solutions role. It signals that you separate what a platform does from how it is marketed, which is exactly the trust a technical buyer is looking for.
Frequently asked questions
Is personalization the same as AI?
No. Most personalization is templating: merge tags and conditionals that fill a template per recipient. AI is an added layer that optimizes timing, channel, and audience, or generates copy.
What is the difference between predictive and generative AI here?
Predictive AI forecasts a value or choice from behavior, such as the best send time or channel. Generative AI writes new content, such as a subject line. They solve different problems and should not be conflated.
What are merge tags?
Placeholders in a template that get filled in per recipient at send time, such as a first name or cart contents. They render a different message per person from one template. Iterable implements them with Handlebars syntax, which also supports conditional logic.
Keep reading
Sources
Part of the MarTech and customer engagement field notes on this site. Written by Antoni K Pestka.