Skip to content

indranilbanerjee/digital-marketing-pro / skills/context-engine

context-engine

Verified

Load shared marketing context: brand profile, benchmarks, compliance rules, specs. "load my brand context"

FileKindSize
30-60-90-framework.mdmarkdown8.2 KB
SKILL.mdskill md12.9 KB
actionable-persona-format.mdmarkdown7.0 KB
advanced-reporting-guide.mdmarkdown21.1 KB
agency-operations-guide.mdmarkdown15.7 KB
ai-visibility-metrics.mdmarkdown2.9 KB
approval-framework.mdmarkdown13.0 KB
b2b-decision-making-unit.mdmarkdown9.1 KB
channel-families.mdmarkdown7.7 KB
competitive-monitoring-guide.mdmarkdown16.8 KB
competitor-3-question-output.mdmarkdown8.0 KB
compliance-rules.mdmarkdown87.2 KB
compound-intelligence-guide.mdmarkdown15.5 KB
creative-intelligence-guide.mdmarkdown13.9 KB
crm-integration-guide.mdmarkdown35.4 KB
custom-mcp-guide.mdmarkdown17.1 KB
decision-framework.mdmarkdown7.4 KB
decision-matrix-rerun.mdmarkdown4.9 KB
engagement-flow-methodology.mdmarkdown11.4 KB
eu-code-of-practice.mdmarkdown15.6 KB
eval-framework-guide.mdmarkdown27.5 KB
eval-rubrics.mdmarkdown16.8 KB
execution-workflows.mdmarkdown23.3 KB
five-digital-markets.mdmarkdown7.5 KB
fixed-vs-variable-budget.mdmarkdown6.8 KB
four-core-documents-spec.mdmarkdown14.2 KB
geo-execution-guide.mdmarkdown20.2 KB
google-seo-reference.mdmarkdown7.3 KB
growth-plan-template.mdmarkdown9.5 KB
guidelines-framework.mdmarkdown14.0 KB
in-market-out-market.mdmarkdown6.1 KB
india-market-context.mdmarkdown14.1 KB
industry-profiles.mdmarkdown65.0 KB
intelligence-layer.mdmarkdown3.9 KB
journey-growth-guide.mdmarkdown19.8 KB
living-instruction-file-spec.mdmarkdown7.2 KB
market-intelligence-guide.mdmarkdown13.7 KB
marketing-science-guide.mdmarkdown14.0 KB
memory-architecture.mdmarkdown12.5 KB
monthly-report-template.mdmarkdown7.0 KB
multilingual-execution-guide.mdmarkdown31.7 KB
narrative-warfare-guide.mdmarkdown17.9 KB
platform-publishing-specs.mdmarkdown19.5 KB
platform-specs.mdmarkdown34.7 KB
reporting-cadence.mdmarkdown8.5 KB
schema-templates.jsonconfig8.7 KB
scoring-rubrics.mdmarkdown10.9 KB
self-healing-ops-guide.mdmarkdown18.5 KB
seo-execution-guide.mdmarkdown14.8 KB
stone-vs-opinion.mdmarkdown7.7 KB
synthetic-audience-guide.mdmarkdown20.4 KB
team-roles-framework.mdmarkdown12.4 KB
three-scenario-forecasting.mdmarkdown7.4 KB
transcreation-framework.mdmarkdown24.2 KB
two-views-model.mdmarkdown5.5 KB
unit-economics-framework.mdmarkdown7.5 KB
update-back-rule.mdmarkdown7.0 KB
yearly-planner-template.mdmarkdown6.1 KB

synthetic-audience-guide.md

View on GitHub
# Synthetic Audience Testing Guide

Reference knowledge for AI-simulated audience research, synthetic persona construction, focus group simulation, message testing, pricing sensitivity estimation, and calibration methodology. Use this to pre-screen marketing decisions before committing real budget, while maintaining rigorous awareness of the method's limitations.

---

## 1. Synthetic Persona Construction from CRM Data

### Data Extraction Process
Build personas from actual behavioral data, not assumptions. Pull the following from your CRM and analytics platforms:

**Behavioral Clusters**: Segment customers by observable behavior patterns:
- **Purchase patterns**: Frequency (weekly, monthly, quarterly, annual), average order value, category breadth (single-category vs. multi-category), purchase timing (beginning of month, end of quarter, seasonal)
- **Engagement levels**: Email open/click rates, website visit frequency, content downloads, event attendance, support interactions, product usage depth
- **Acquisition channel**: Organic search, paid search, social, referral, direct, partner, event. Channel of origin correlates with expectations and behavior
- **Lifetime value tier**: Top 10% (power users), 11-30% (engaged), 31-70% (moderate), 71-100% (low-value/at-risk). Behavior and motivations differ significantly across tiers

**Demographic Foundation**: For each behavioral cluster, extract the demographic profile:
- Age range (not single age — clusters span ranges)
- Geographic distribution (regions, urban/suburban/rural)
- Industry and company size (B2B) or income bracket (B2C)
- Job title/role (B2B) or household composition (B2C)

**Psychographic Layer**: Derived from qualitative data sources:
- **Support tickets**: What do they complain about? What language do they use? What do they praise?
- **Survey responses**: NPS comments, satisfaction survey freeform text, post-purchase feedback
- **Review text**: G2, Capterra, Trustpilot, Amazon reviews — actual words customers use to describe their experience
- **Sales call notes**: Objections raised, questions asked, priorities stated, decision criteria mentioned
- **Community posts**: Reddit discussions, Slack community messages, forum posts about your product or category

### Persona Template

```
Persona: [Cluster Name]
Segment size: [% of customer base]
Behavioral signature: [2-3 sentence summary of defining behaviors]

Demographics:
  Age range: [e.g., 30-45]
  Location: [e.g., US metro areas, tier 1 cities]
  Role/title: [e.g., Marketing Manager, Director level]
  Company size: [e.g., 50-500 employees]

Psychographics:
  Primary motivation: [What drives their purchase? Efficiency, growth, risk reduction, status]
  Core objection: [What holds them back? Price, complexity, switching cost, trust]
  Decision style: [Analytical/data-driven, consensus/committee, impulse/speed-oriented, relationship-driven]
  Information sources: [Where do they research? Peers, analysts, reviews, social, communities]
  Language patterns: [Actual phrases from reviews/tickets — "easy to use," "saves me hours," "too expensive for what it does"]

Behavioral patterns:
  Purchase frequency: [Quantified]
  Average order value: [Range]
  Preferred channels: [Email, social, in-app, SMS]
  Content preferences: [Case studies, how-to guides, video demos, data reports]
  Churn risk factors: [What predicts disengagement for this cluster?]
```

### Critical Rule: Ground in Real Data
Every attribute in the persona must be traceable to actual CRM data, survey responses, or customer verbatims. If you cannot cite a data source for an attribute, mark it as an assumption and prioritize validating it. Synthetic personas are only as valuable as the real data that underlies them.

---

## 2. Focus Group Simulation Methodology

### Panel Construction
1. Select 8-12 synthetic personas representing your target segments. Weight the panel to mirror your actual customer composition (if 40% of customers are in Cluster A, 40% of panelists should be Cluster A personas)
2. Include at least 1-2 personas representing segments you want to grow into (aspirational targets). Flag these as aspirational so their responses are weighted appropriately
3. Include 1 "skeptic" persona — someone who has evaluated but not purchased, or a lapsed customer. Their objections reveal blind spots in your messaging

### Stimulus Presentation
Present the material being tested to each persona with context appropriate to how they would actually encounter it:
- **Ad creative**: "You're scrolling through [platform] during [time context]. You see this ad. What is your reaction?"
- **Pricing page**: "You've been researching [category] solutions for 2 weeks. You visit this pricing page. What are you thinking?"
- **Email subject line**: "You receive this email on a Tuesday morning. Your inbox has 47 unread messages. Do you open this?"
- **Product positioning**: "A colleague describes this product to you in one sentence: [positioning statement]. What do you think?"

### Response Generation
For each persona, generate responses that reflect their documented:
- Decision style (analytical personas want data, relationship-driven personas want references)
- Language patterns (use the actual vocabulary from their cluster's reviews and tickets)
- Core objections (their default resistance to new messaging)
- Motivation alignment (does this stimulus speak to their primary motivation or miss it?)

### Analysis Framework
After generating responses from all panelists:

| Analysis | What to Look For |
|---|---|
| **Consensus themes** | Responses that are consistent across 6+ of 8 personas regardless of segment. High-confidence finding |
| **Segment divergence** | Where Cluster A loves it but Cluster B resists. Indicates the need for segment-specific messaging |
| **Objection patterns** | Common objections across personas. Prioritize addressing the most frequent objection |
| **Enthusiasm signals** | Responses with strong positive language from high-value personas. Validate that your best customers resonate |
| **Confusion signals** | Requests for clarification or misinterpretation of the message. Indicates clarity problem, not a persuasion problem |

---

## 3. Message Testing Protocols

### A/B Message Variant Testing
Present 2-4 message variants to the full synthetic panel. Score each variant on five dimensions per persona:

| Dimension | Score (1-10) | Question the Persona Answers |
|---|---|---|
| **Resonance** | How much does this speak to my specific needs? | "Does this feel like it was written for someone like me?" |
| **Clarity** | Do I understand what is being offered and what I should do next? | "Can I explain this to a colleague in one sentence?" |
| **Credibility** | Do I believe this claim based on what I know about the brand? | "Is there evidence to support this, or is it just marketing?" |
| **Urgency** | Am I motivated to act now rather than later? | "Is there a reason I shouldn't just bookmark this and come back later?" |
| **Differentiation** | Is this meaningfully different from alternatives I have seen? | "Could a competitor say the exact same thing?" |

### Scoring and Ranking
1. Calculate average score per dimension per variant across all personas
2. Calculate overall variant score: weighted average of all five dimensions. Default weights: Resonance 25%, Clarity 20%, Credibility 25%, Urgency 15%, Differentiation 15%. Adjust weights based on campaign objective (awareness campaigns weight Differentiation higher; conversion campaigns weight Urgency higher)
3. Identify the **overall winner** (highest aggregate score) and the **segment winner** for each priority segment (highest score among personas in that segment)
4. If the overall winner and segment winner differ, consider segment-specific messaging rather than a single universal message

### Headline and Subject Line Pre-Screening
For high-volume testing (screening 10+ subject line variants to narrow to 3 for real A/B testing):
- Score each variant on Resonance and Clarity only (speed over depth)
- Rank by composite score
- Advance the top 3 to real-world A/B testing with actual audience
- This reduces real-world testing cost by 60-70% while maintaining quality

---

## 4. Pricing Sensitivity Estimation

### Van Westendorp Price Sensitivity Meter (Simulated)
For each persona, estimate four price thresholds for the product or service:

| Threshold | Question | What It Reveals |
|---|---|---|
| **Too cheap** | "Below what price would you question the quality?" | Floor of credible pricing. Pricing below this triggers quality concern |
| **Cheap (good deal)** | "At what price would you feel you're getting a good deal?" | Sweet spot for value-conscious segments. Promotional pricing target |
| **Expensive (but would consider)** | "At what price would you start to hesitate but still consider?" | Upper boundary of consideration. Where price objection begins |
| **Too expensive** | "Above what price would you not consider this product regardless of features?" | Ceiling. Pricing above this loses the segment entirely |

### Aggregate Analysis
Plot the four price curves across all personas:
- **Optimal Price Point (OPP)**: Intersection of "too cheap" and "too expensive" curves. The price where the fewest people object on either end
- **Indifference Price Point (IDP)**: Intersection of "cheap" and "expensive" curves. The price where equal numbers find it cheap vs expensive
- **Acceptable Price Range**: Between the Point of Marginal Cheapness (intersection of "too cheap" and "expensive") and the Point of Marginal Expensiveness (intersection of "cheap" and "too expensive"). Pricing within this range is defensible
- **Segment-specific ranges**: Run the analysis per persona cluster. High-LTV segments typically have wider acceptable ranges and higher OPPs. Price-sensitive segments have narrow ranges and low OPPs. This informs tiered pricing strategy

### Price Elasticity Estimation
For each persona cluster, estimate how purchase probability changes with price:
- At OPP: ~80% purchase probability
- At IDP + 10%: ~60% purchase probability
- At "expensive" threshold: ~40% purchase probability
- At "too expensive" threshold: ~5% purchase probability

Use these estimates to model revenue at different price points: `Revenue = Price * Purchase_probability * Segment_size`. The revenue-maximizing price is typically above the OPP (you sacrifice some volume for higher margin).

---

## 5. Statistical Validity and Limitations

### What Synthetic Testing IS
- **Directional guidance**: Points you toward the right answer, not the exact answer. "Message A is likely stronger than Message B for this segment"
- **Hypothesis generation**: Creates testable hypotheses for real-world experimentation. "We hypothesize that the value-framed message will outperform the fear-framed message among mid-market prospects"
- **Pre-screening filter**: Eliminates clearly weak options before real testing. Narrows from 10 variants to 3, saving real-world testing budget
- **Rapid iteration tool**: Test 20 variants in minutes rather than weeks. Useful for early-stage ideation where speed matters more than precision
- **Scenario planning**: Model how different segments might respond to competitor moves, pricing changes, or messaging shifts before they happen

### What Synthetic Testing IS NOT
- **Replacement for real customer research**: Synthetic responses are modeled projections, not observed behavior. High-stakes decisions (product launches, rebrand, major pricing changes) require real-world validation
- **Exact quantitative predictions**: "72% of synthetic panelists preferred Message A" does NOT mean 72% of real customers will prefer it. Treat percentages as relative rankings, not absolute predictions
- **Validation of novel ideas**: Synthetic personas respond based on historical patterns. Truly novel products or positioning may not be accurately modeled by personas built on past behavior
- **Substitute for talking to customers**: Synthetic testing supplements human research. It does not replace it. Use synthetic testing between research cycles, not instead of them

### Confidence Level Guide

| Scenario | Synthetic Confidence | Action |
|---|---|---|
| Screening 10 subject lines to find 3 for real testing | High — synthetic testing is excellent at eliminating weak options | Use results directly to narrow the field |
| Choosing between 2 positioning statements for a rebrand | Low — high-stakes decision with novel elements | Use to form hypothesis, then validate with real customer research |
| Estimating pricing sensitivity for a new product tier | Medium — directional, but real market behavior may diverge | Use to set initial hypothesis, then run real Van Westendorp with 200+ respondents |
| Pre-screening ad creative concepts before production | Medium-high — good at identifying resonance and clarity issues | Use to eliminate weak concepts, produce top 3-4 for real testing |
| Modeling response to a competitor's price cut | Medium — useful for scenario planning, not prediction | Use to prepare response options, then monitor real market response |

---

## 6. Bias Mitigation

### Confirmation Bias
**Risk**: Tuning persona responses to confirm the outcome you already prefer. If you want Message A to win, the personas magically prefer Message A.
**Mitigation**: Define persona response parameters BEFORE generating responses to any stimulus. Lock the persona profiles, then present the stimulus. Have a second reviewer audit responses for consistency with the persona profile. If a price-sensitive persona suddenly does not object to premium pricing, the response is suspect.

### Representation Bias
**Risk**: Building personas that represent your ideal customers rather than your actual customer base. The panel skews toward high-LTV, highly engaged customers because those are the most visible in your data.
**Mitigation**: Weight the panel to match actual customer composition. If 60% of your revenue comes from mid-market companies, 60% of the panel should be mid-market personas. Include low-engagement segments — they are often the majority of your base and the hardest to retain.

### Survivorship Bias
**Risk**: Only modeling current happy customers. Ignoring churned customers, lost prospects, and people who evaluated but did not buy.
**Mitigation**: Include at least 2 personas built from churned customer data (exit surveys, cancellation reasons, pre-churn behavior patterns) and lost-deal data (CRM lost-deal reasons, competitor chosen, objections cited). These personas reveal messaging weaknesses that your current customer personas will never surface.

### Recency Bias
**Risk**: Over-indexing on recent customer behavior at the expense of long-term patterns. A seasonal spike becomes a permanent persona trait. A temporary market condition shapes a persona that does not reflect the norm.
**Mitigation**: Build personas from at least 12 months of data to smooth seasonal effects. Flag any persona attribute that is based on less than 6 months of data as potentially unstable. Compare persona profiles year-over-year to identify attributes that are durable versus transient.

### Language Bias
**Risk**: Personas use marketing language ("synergize," "leverage," "drive ROI") instead of customer language ("makes my job easier," "saves me from Saturday work," "my boss finally sees the numbers").
**Mitigation**: Source all persona language directly from customer verbatims: review text, support tickets, survey freeform fields, sales call transcripts, community posts. If a persona's response sounds like a marketing brief, it is not grounded in real customer expression.

---

## 7. Calibration Against Real-World Results

### Calibration Loop
After every real campaign, research study, or A/B test:
1. Pull the synthetic prediction made before the real test (what did the synthetic panel predict?)
2. Record the actual outcome (what happened in reality?)
3. Calculate prediction accuracy: `accuracy = 1 - abs(predicted_rank - actual_rank) / number_of_variants`. For directional predictions (which variant wins), score binary: correct or incorrect
4. Log accuracy by persona cluster. Which personas are well-calibrated? Which consistently over-predict or under-predict?
5. Update the persona model for poorly calibrated clusters. Adjust attributes, language, or response tendencies based on the delta between prediction and reality

### Calibration Scorecard

| Metric | Calculation | Target | Action if Below Target |
|---|---|---|---|
| **Directional accuracy** | % of times synthetic panel correctly predicted the winning variant | >70% | Review persona profiles for staleness, bias, or missing segments |
| **Ranking accuracy** | Average position delta between synthetic ranking and real ranking (for 3+ variants) | <1.0 position delta | Adjust persona scoring weights or add missing personas |
| **Segment accuracy** | % of times segment-specific synthetic prediction matched segment-specific real result | >60% | Refresh segment personas with fresh CRM data and new verbatims |
| **Pricing accuracy** | Delta between synthetic OPP and real-world optimal price point | <15% delta | Re-ground pricing thresholds with fresh purchase data and real Van Westendorp |

### Quarterly Refresh Cycle
1. Pull fresh CRM data for all persona clusters (new purchase patterns, updated engagement metrics, recent survey responses)
2. Re-run cluster analysis to check if segments have shifted (new clusters emerging, existing clusters merging)
3. Update persona demographics with current data
4. Replace stale verbatims with recent customer language (from last 90 days of reviews, tickets, and surveys)
5. Add new personas for any new segments (new market entries, new product lines, new geographic expansion)
6. Retire personas for segments you have exited or that no longer represent meaningful customer volume
7. Document all changes with date and rationale for audit trail

---

## 8. Use Cases and Application Patterns

### Pre-Launch Product Messaging
Test positioning variants before spending on ads. Run 4-6 positioning statements through the synthetic panel. Identify which framing resonates most with your priority segments. Use the winner as the foundation for ad creative, landing pages, and sales enablement. Confidence: medium-high for eliminating weak options, medium for predicting the absolute winner.

### Pricing Strategy Development
Before real market exposure, run pricing scenarios through the synthetic panel. Identify the acceptable price range per segment. Model revenue at different price points. Use as the starting hypothesis for real-world pricing tests (free trial conversion rates, upgrade rates, price page A/B tests).

### Competitive Response Planning
When a competitor makes a move (price cut, feature launch, rebrand), quickly assess how your audience might respond. Which segments are most vulnerable to the competitor's new positioning? Which counter-narrative is most likely to resonate? Use to prepare response options within 48 hours rather than weeks.

### Content Strategy Prioritization
Test topic angles and content formats before production. Present 5-8 content concepts to the synthetic panel. Score on relevance, share-worthiness, and information need. Prioritize production of the highest-scoring concepts. Reduces content waste from producing pieces that miss the audience's actual interests.

### Email Subject Line Pre-Screening
Generate 10-15 subject line variants. Run through synthetic panel scoring for open probability (based on Resonance and Clarity). Advance top 3 to real-world A/B testing. This two-stage approach (synthetic screening then real testing) narrows the field cheaply and focuses real testing budget on the strongest contenders.

### Crisis Messaging Development
When a crisis hits (product issue, PR incident, negative press), test response messaging options before publishing. Run 3-4 response statements through the synthetic panel. Identify which response best addresses customer concerns per segment. Identify which response minimizes negative perception. Deploy the synthetic-validated response while preparing real-time monitoring to adjust if actual response diverges from the prediction.

### New Market Entry Assessment
Before entering a new geographic market or industry vertical, build synthetic personas from available data (industry reports, competitor customer reviews in that market, publicly available survey data). Test your current messaging against these personas to identify necessary adaptations. Flag areas where your current positioning may not translate and where localized messaging is required. Confidence is lower for new markets (less CRM data to ground the personas) — weight real research more heavily.
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 194kUpdated Claude CodeCodexVerifiedInstall