AI SEGMENTATION APPROACHES
Segmentation Methods
| Method |
Description |
Best For |
| RFM Analysis |
Recency, Frequency, Monetary |
Transaction-based |
| Clustering |
K-means, hierarchical |
Pattern discovery |
| Predictive |
ML-based scoring |
Likelihood to convert |
| Behavioral |
Action-based grouping |
Product engagement |
| Lookalike |
Similar to best customers |
Acquisition |
Method Selection
| Scenario |
Recommended Method |
| Ecommerce customers |
RFM + Clustering |
| SaaS users |
Behavioral + Predictive |
| Lead prioritization |
Predictive scoring |
| New customer acquisition |
Lookalike |
| Content personalization |
Behavioral |
DATA REQUIREMENTS
Data Points Needed
| Category |
Data Points |
Source |
| Demographic |
Title, company, industry |
CRM, enrichment |
| Firmographic |
Size, revenue, location |
CRM, enrichment |
| Behavioral |
Pages viewed, features used |
Analytics, product |
| Transactional |
Purchases, revenue |
CRM, billing |
| Engagement |
Opens, clicks, responses |
Marketing tools |
Data Quality Requirements
| Requirement |
Why |
Minimum Standard |
| Completeness |
Accurate clustering |
>80% fill rate |
| Accuracy |
Valid segments |
Verified sources |
| Recency |
Current behavior |
<90 days |
| Volume |
Statistical validity |
1,000+ records |
MODEL BUILDING
Clustering Process
1. Select variables
2. Normalize data
3. Choose algorithm (K-means typical)
4. Determine optimal K (elbow method)
5. Run clustering
6. Validate segments
7. Name and describe segments
AI Segmentation Prompt
Analyze this customer data and identify distinct segments:
Data fields:
[list available fields]
Sample data:
[paste sample]
For each segment provide:
- Defining characteristics
- Size (% of total)
- Value (revenue/LTV)
- Recommended messaging
- Channel preferences
MODEL VALIDATION
Validation Criteria
| Criterion |
Question |
Threshold |
| Size |
Is each segment large enough? |
>5% of base |
| Distinct |
Are segments different? |
Clear separation |
| Stable |
Does it hold over time? |
>80% consistency |
| Actionable |
Can we reach them differently? |
Yes |
| Valuable |
Do they respond differently? |
>10% lift |
Segment Validation Table
| Segment |
Size |
Avg. LTV |
Engagement |
Valid? |
| Segment A |
% |
$ |
|
|
| Segment B |
% |
$ |
|
|
| Segment C |
% |
$ |
|
|
SEGMENT ACTIVATION
Activation Strategies
| Segment Type |
Activation |
| High-value |
VIP treatment, upsell |
| At-risk |
Retention campaigns |
| New |
Onboarding, education |
| Low-engagement |
Re-engagement |
| High-potential |
Nurture to conversion |
Channel Mapping
| Segment |
Email |
Ads |
Sales |
Content |
| Enterprise |
Low freq |
LinkedIn |
High touch |
Case studies |
| SMB |
Medium freq |
Google |
Self-serve |
How-tos |
DYNAMIC SEGMENTATION
Real-Time Triggers
| Trigger |
Segment Change |
Action |
| First purchase |
Prospect → Customer |
Welcome series |
| 30 days inactive |
Active → At-risk |
Re-engagement |
| Price page visit |
Browsing → Intent |
Sales alert |
| Support ticket |
Any → Needs attention |
Priority support |
PERFORMANCE MONITORING
Segment Health Dashboard
| Segment |
Size |
Size Change |
Conversion |
Revenue |
| Segment A |
|
|
|
|
| Segment B |
|
|
|
|