# Iteration
Your first email generation is a starting point, not a finished campaign. Iteration is how you close the gap between what the AI produces and what your audience actually needs — and in Humanic, it's built directly into the conversation.
***
### What iteration means in Humanic
Unlike traditional email tools where editing means going back into a builder and manually adjusting fields, iteration in Humanic is conversational. You stay in the same chat thread, give the AI feedback or a new instruction, and it regenerates — keeping all the context from your earlier exchanges intact.
This means you don't lose your work. Every correction you make teaches the AI more about your preferences, making each subsequent output sharper than the last.
***
### When to iterate
You should iterate whenever the output is close but not quite right. Common scenarios include:
The tone feels off — too formal, too casual, or not aligned with your brand voice.
The subject line isn't compelling enough for your audience.
The call to action is too vague or doesn't match the goal of the campaign.
The email is too long or too short for the context.
The messaging doesn't reflect the specific micro-cohort you're targeting.
The offer or value proposition needs to be repositioned.
You don't need to regenerate from scratch in any of these cases. A single follow-up instruction is usually enough.
***
### How to iterate effectively
**Be specific about what to change**
Vague feedback produces vague results. Instead of "make it better," say "shorten the email to three paragraphs and make the CTA more urgent." The AI responds to precise instructions the same way a copywriter would.
**Tell the AI what to keep**
If 80% of the email is working, say so. For example: "Keep the subject line and the opening paragraph, but rewrite the body to focus more on the time-limited offer." This prevents the AI from regenerating content you're already happy with.
**Change one thing at a time**
If you ask the AI to change the tone, the structure, and the offer all in one message, the output can drift in unpredictable ways. Make targeted, sequential changes — especially when you're close to a final version.
**Use the conversation as a log**
Everything you've said in the thread shapes the AI's understanding of your campaign. If you gave context early on — the audience, the product, the goal — the AI carries that forward. You don't need to repeat yourself each time you iterate.
***
### Types of iteration
**Content iteration**
Changing what the email says — the subject line, opening hook, body copy, CTA, or sign-off. This is the most common type and usually takes one to two exchanges to get right.
**Tone iteration**
Adjusting how the email sounds — more conversational, more authoritative, more empathetic, more urgent. Tone changes often benefit from a short example: "Write it in a tone similar to this: \[paste example]."
**Structure iteration**
Changing how the email is organized — moving the CTA higher, breaking one long paragraph into three short ones, adding a bullet list for clarity, or removing a section that buries the main message.
**Audience iteration**
Refining who the email speaks to. If you started with a broad cohort and want to tighten the message for a specific segment — for example, users who completed onboarding but haven't used a key feature — describe the new audience and ask Humanic to re-tailor the email accordingly.
**Offer iteration**
Adjusting the incentive, discount, feature highlight, or value proposition the email leads with. This is especially useful when A/B testing different angles for the same campaign goal.
***
### How many iterations does it typically take?
Most campaigns reach a publish-ready state in two to three exchanges. The first generation establishes the foundation. The second exchange refines tone and structure. The third, if needed, sharpens the CTA or adjusts the opening.
If you find yourself iterating more than four or five times on the same email, it usually means the original prompt lacked enough context. Go back and add more detail about the audience, goal, or constraints — then regenerate from a stronger starting point.
***
### Iteration and brand learning
Every correction you make in a conversation contributes to Humanic's understanding of your brand preferences — especially if you have "Keep refining based on conversations" enabled in your Campaign Preferences.
Over time, this means fewer iterations per campaign. The AI starts anticipating your preferences and producing first drafts that are already closer to your standards.
***
### What iteration is not
Iteration is not the same as starting over. If you find yourself discarding the entire output and rewriting the prompt from scratch every time, that's a signal to invest more time upfront in your brand guidelines and prompt structure — not to iterate more aggressively.
Iteration works best as a refinement tool, not a replacement for a clear initial brief.
***
### Frequently asked questions
**Does iterating in a conversation affect other campaigns?**
No. Iterations within a conversation only apply to that session. Broader learning happens through your Campaign Preferences, which update separately based on your overall usage patterns.
**Can I iterate after a campaign has been sent?**
You can use insights from a sent campaign — open rates, click-through rates, replies — to inform the prompt for your next campaign. Humanic's analytics surface what worked, and you can feed those learnings directly into your next iteration cycle.
**What if the AI keeps repeating the same mistake?**
If the AI misunderstands a specific instruction repeatedly, try rephrasing it with an example or a negative constraint. For instance: "Do not open with a question" is more reliable than "make the opening more direct."
**Is there a limit to how many times I can iterate?**
No. You can exchange as many messages as needed within a conversation. There is no generation limit per session.
---
# Agent Instructions: Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://humanic.gitbook.io/humanic/advanced-features/hyper-personalization/iteration.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.