Using AI for Email Personalisation: Tools, Techniques and Practical Implementation
By Email ExtractorPublished 6 min read
On this page
Why AI Changes Email Personalisation
Traditional email personalisation relies on merge fields ({first_name}, {company}) and segment-based content. AI enables personalisation at the individual level by generating unique content for each recipient:
Personalisation level
Traditional approach
AI-enhanced approach
Name and company
Merge fields: "Hi {first_name}"
Same (AI not needed for this)
Industry
Segment-based templates per industry
AI writes industry-specific opening lines per recipient
Role-based messaging
3-5 templates per persona
AI adjusts tone, pain points and value prop per individual role
Company-specific reference
Manual research per prospect (does not scale)
AI reads company website, news and LinkedIn to write custom lines
Behavioural context
"You downloaded our whitepaper"
AI interprets behaviour pattern and adjusts entire email
Competitive positioning
Generic competitive comparisons
AI references their specific current tools and tailors positioning
Timing context
Day/time optimisation
AI references their recent activity, news, or trigger events
Where AI personalisation has the most impact
Email type
Personalisation impact
AI uplift
Notes
Cold outreach (first touch)
Very high
2-3x reply rate improvement
Opening line and relevance are critical
Follow-up sequences
High
1.5-2x reply rate improvement
AI adjusts based on non-response context
Re-engagement campaigns
Medium-high
1.5-2x open rate improvement
AI references past interaction history
Newsletter content
Medium
1.2-1.5x click rate improvement
AI selects and orders content per subscriber
Transactional emails
Low
Minimal
Standardised content is expected
Customer success check-ins
Medium-high
1.5-2x response rate
AI references product usage and account context
AI Personalisation Techniques
Opening line generation
The opening line is the highest-leverage personalisation point. It determines whether the recipient reads the rest:
Input data
AI-generated opening line example
Quality
Company website
"Your approach to [specific product feature] at [company] caught my attention, especially [detail]"
High (specific and relevant)
Recent press release
"Congratulations on [specific announcement]. The timing is relevant because..."
High (timely and personal)
LinkedIn profile
"Your background in [specific experience] makes you uniquely positioned for [topic]"
Medium-high (personal but common)
Job posting
"I noticed [company] is building out the [team] function. That growth stage typically means..."
High (intent-driven)
Conference speaker
"Your [conference] talk on [topic] resonated, particularly [specific point]"
High (flattering and specific)
No specific data (generic)
"As a [title] at [company], you likely deal with [generic pain point]"
Low (template-feeling)
Full email generation vs hybrid approach
Approach
How it works
Pros
Cons
Full AI generation
AI writes entire email from scratch per recipient
Maximum personalisation; unique per person
Quality inconsistent; harder to control; may not match brand voice
Hybrid (AI opening + human template)
AI writes personalised first 1-2 lines; rest is tested template
Personalised hook; consistent value prop; quality controlled
Opening-body transition can feel jarring
AI-selected content blocks
AI selects from pre-written content blocks per recipient
Instruct AI to be specific rather than superlative
Outdated reference
References role or company from 2 years ago
Date validation on source data
Use current data only; timestamp enrichment data
Quality scoring rubric
Criterion
Score 1 (reject)
Score 3 (acceptable)
Score 5 (excellent)
Accuracy
Contains factual errors
All facts correct; some generic
All facts correct and specific
Relevance
No connection to recipient's context
Relevant to their industry or role
Specific to their company and situation
Tone
Wrong tone for audience
Appropriate tone; slightly template-like
Natural; sounds like a knowledgeable person wrote it
Length
Too long or too short
Appropriate length
Optimal length for the channel
Call to action
No CTA or aggressive CTA
Clear CTA; somewhat generic
CTA tied to their specific situation
Naturalness
Reads like AI generated it
Mostly natural; minor AI tells
Reads like a human wrote it
Metrics
Metric
Non-personalised cold email
Basic personalisation (merge fields)
AI-personalised cold email
Open rate
20-30%
25-35%
35-50%
Reply rate
1-3%
3-5%
5-12%
Positive reply rate
0.5-1.5%
1-3%
3-8%
Meeting booked rate
0.3-1%
0.5-2%
2-5%
Spam complaint rate
0.1-0.3%
0.05-0.15%
0.02-0.1%
Unsubscribe rate
0.5-2%
0.3-1%
0.1-0.5%
Preparing Prospect Data for AI Personalisation
Before feeding prospect data into AI personalisation tools, you need clean, deduplicated prospect lists. When compiling prospect data from CRM exports (CSV), LinkedIn research, web scraping (HTML, JSON), enrichment tool exports, conference attendee lists (PDF) and purchased data lists, upload the files to Email Extractor to extract and deduplicate email addresses across all sources. AI personalisation tools charge per email generated, so sending duplicate prospects through the personalisation pipeline wastes API credits and risks sending the same prospect two different personalised emails, which immediately signals automation.
