Personalization has three levels, and they are not interchangeable. Inserting a first name is a formatting step. Referencing the prospect’s company situation is context. Saying something about that situation the recipient has not heard from a competitor is insight. Most "personalized" outreach stops at level one, which is why reply rates do not move.
Each level costs more to produce and is harder to fake. The returns are not linear — level two is where the step change happens.
| Level | What it uses | Typical result |
|---|---|---|
| 1 — Merge fields | First name, company name, job title | Reads as automated on the first line |
| 2 — Context | Industry, company size, role, public signals about the company | Reads as researched; clears the relevance bar |
| 3 — Insight | A specific observation about their situation and a matching implication | Earns a reply even from a cold contact |
Because recipients pattern-match. A greeting with a familiar first name inside an otherwise generic message is recognised as a template within one sentence, and the recognition transfers to everything after it. The problem is not that merge fields are dishonest — it is that they do not carry information the recipient values.
Writing a level-two email by hand takes several minutes of research per prospect, which does not scale past a few dozen contacts a week. A model can produce level-two copy per prospect from context that is already in the lead record — which is what makes personalization sustainable at volume rather than a heroic first week.
Not by reading the emails. Compare reply rate between a segmented and an unsegmented send of the same offer, and watch the distribution rather than the average — personalization usually shows up as a shift in which accounts reply, not just how many.
Mailor generates each email from the prospect’s lead record — company, role and the ICP the lead was scored against — rather than from a template. Because ICP inference runs first, personalization has a definition of relevance to work from. Emails come out in six tones (direct, value-first, curiosity, social-proof, story, or auto) and 18 languages with a spam pre-check, and every draft is editable and regenerable.
Technically it is a merge field, which is level one and the least effective. It signals that automation is in play without supplying anything the recipient values. Context about the company or role is what actually moves reply rates.
That is its main advantage. Hand-researched copy does not scale past a few dozen contacts a week; generating per-prospect copy from context already stored on the lead record scales to the full list without falling back to templates.
Enough to match segment and geography without rewriting by hand. Mailor offers six tones (direct, value-first, curiosity, social-proof, story, or auto) and 18 languages, with length guardrails that keep emails inside the range cold-email practice consistently favours.
Yes, and it should be measured by segment rather than on average. Send the same offer to two segments with and without context-aware copy and compare reply rate; if the numbers match, the personalization is not doing any work.