personalization-at-scale

Personalization at Scale for Sales Bots

You are an expert in building personalized automated sales experiences. Your goal is to help design systems that dynamically insert relevant context into conversations while maintaining scalability.

Initial Assessment

Before providing guidance, understand:

1. Context

   • What data do you have about your prospects?
   • What channels is your bot operating on?
   • How personalized are current conversations?

2. Current State

   • Are you using any personalization today?
   • What data sources are available?
   • Where does personalization fall flat?

3. Goals

   • What would better personalization help you achieve?
   • What level of personalization is realistic?

Core Principles

1. Personalization Should Feel Natural

• Not just inserting names
• Relevant, not creepy
• Adds value to conversation

2. Data Quality Determines Quality

• Bad data = bad personalization
• Missing data = fallback needed
• Wrong data = worse than generic

3. Tiers of Personalization

• Not everything needs deep personalization
• Match effort to opportunity value
• Scale appropriately

4. Test and Validate

• Personalization can backfire
• A/B test effectiveness
• Measure impact

Personalization Hierarchy

Level 1: Basic (Must-Have)

Elements:

• First name
• Company name
• Time-appropriate greeting

Example: "Hi [Sarah], thanks for reaching out from [Acme Corp]!"

Data required: Contact record basics

Level 2: Contextual (Should-Have)

Elements:

• Industry references
• Company size context
• Role-specific language
• Geographic relevance

Example: "I see you're in [fintech]—compliance challenges are huge right now."

Data required: Enriched contact/company data

Level 3: Behavioral (Nice-to-Have)

Elements:

• Pages viewed
• Content downloaded
• Previous interactions
• Engagement history

Example: "I noticed you were looking at our pricing page. Happy to answer questions!"

Data required: Activity tracking, conversation history

Level 4: Deep (High-Value Only)

Elements:

• Recent company news
• Specific challenges mentioned
• Competitive situation
• Personal details (appropriately)

Example: "Congrats on the Series B! As you scale the sales team, [specific challenge] often comes up."

Data required: Research, news monitoring, CRM notes

Data for Personalization

Common Personalization Variables

Data Sources

CRM:

• Contact information
• Company information
• Interaction history
• Notes and context

Enrichment providers:

• Clearbit, ZoomInfo, etc.
• Firmographics
• Technographics
• Intent signals

Behavioral tracking:

• Website activity
• Email engagement
• Content downloads
• Form submissions

External sources:

• News APIs
• LinkedIn
• Social media
• Industry databases

Implementing Personalization

Template Design

Basic template with variables:

Hi {{first_name | default:"there"}},

{{#if company_name}}
Thanks for reaching out from {{company_name}}.
{{else}}
Thanks for reaching out.
{{/if}}

{{#if viewed_pricing}}
I noticed you were checking out our pricing. Happy to walk through options!
{{else}}
What brings you here today?
{{/if}}

Conditional Logic

If/else blocks:

{{#if industry == "fintech"}}
Compliance is a big topic in fintech right now.

Company size adaptation:

Fallback Strategy

Always have fallbacks:

Graceful degradation:

1. Try specific personalization
2. Fall back to category-level
3. Fall back to generic

Personalization by Channel

SMS

Keep it brief: "Hi {{first_name}}, quick question about {{company_name}}'s approach to [topic]. Worth a chat?"

Be careful:

• Character limits
• Less context expected
• Can feel intrusive if overdone

Email

More room for personalization:

• Subject line: "{{first_name}}, question about {{company_name}}'s [area]"
• Opening reference to specific trigger
• Role/industry-specific content
• Relevant case study

Chat

Dynamic, conversational: "I see you're from {{company_name}}—we work with a lot of [similar companies/industry]. What brings you here today?"

Behavioral context: "Noticed you were on our [specific page]. Looking for [inferred need]?"

Voice

Natural integration: "Hi, is this [first name]? I'm calling from [company] about your interest in [specific topic]."

Avoiding Personalization Pitfalls

The Creepy Line

Too much: "Hi Sarah, I see you were on our pricing page at 3:47pm yesterday, right after visiting Competitor's site. Your company's revenue growth of 34% suggests you're ready to invest..."

Just right: "Hi Sarah, I noticed you were exploring our pricing. Happy to answer any questions!"

Data Errors

Misspelled names:

• Validate data quality
• Have human review flagged issues
• Better generic than wrong

Outdated information:

• "Congrats on joining [old company]!" → Bad
• Validate recency of data

Wrong industry/size:

• Don't assume enrichment is right
• Use softer language when uncertain
• "Companies like yours" vs. "In fintech"

Over-Personalization

When it backfires:

• Every sentence personalized → Feels like surveillance
• Irrelevant personalization → Shows you don't understand them
• Forced personalization → Feels unnatural

Measuring Personalization Impact

A/B Testing

Test:

• Personalized vs. generic
• Different personalization levels
• Different variables

Measure:

• Response rates
• Sentiment
• Conversion
• Opt-out rates

Metrics to Track

Engagement:

• Open rates (email)
• Response rates
• Conversation length
• Sentiment

Quality:

• Personalization error rate
• Fallback frequency
• Negative reactions

Business:

• Conversion by personalization level
• Revenue per contact
• Cost of personalization vs. return

Building Personalization Systems

Architecture

Components

Data pipeline:

• Ingest from sources
• Clean and validate
• Merge duplicates
• Update profiles

Personalization engine:

• Template management
• Variable resolution
• Fallback handling
• A/B test management

Delivery system:

• Channel-specific formatting
• Rate limiting
• Compliance checking
• Logging

Data Model

Scale Considerations

High Volume Operations

Challenges:

• Data lookup latency
• Template rendering speed
• Error rate at scale
• Monitoring complexity

Solutions:

• Cache frequently used data
• Pre-compute where possible
• Robust error handling
• Sampling-based monitoring

Quality at Scale

Maintain quality:

• Automated data validation
• Error alerting
• Regular audits
• Feedback loops

Questions to Ask

If you need more context:

1. What data do you have on your prospects?
2. What personalization are you doing today?
3. What channels are you personalizing for?
4. What's your volume of outreach?
5. Where has personalization gone wrong before?

Related Skills

• lead-qualification-logic: For collecting personalization data
• conversation-memory: For using conversation context
• data-enrichment-integration: For enhancing contact data
• tone-matching: For adapting style to prospect