wiki-category-research.md12.1 KBView on GitHub # Sales Wiki Category Research
Research on optimal knowledge categories for a sales wiki, based on cognitive science, industry standards, and Cedar's existing extraction structure.
---
## Sources
| Source | Topic | URL |
|--------|-------|-----|
| Allego | Sales Content Taxonomy Guide | https://www.allego.com/learning/guide-to-building-a-high-impact-sales-content-taxonomy/ |
| Gong | Theme Analysis | https://help.gong.io/v1/docs/how-to-create-a-theme-analysis |
| MEDDICC | Framework Documentation | https://meddicc.com/what-is-meddpicc |
| MEDDICC | AI/LLM Guidance | https://meddicc.com/llm-info |
| G2 | Sales Playbook Template | https://learn.g2.com/sales-playbook |
| Salesforce | Sales Playbook Guide | https://www.salesforce.com/blog/sales-playbook/ |
| Pipedrive | Sales Playbook Examples | https://pipedrive.com/en/blog/sales-playbook |
| Klue | Competitive Battlecards 101 | https://klue.com/blog/competitive-battlecards-101 |
| Seeto | Battle Card Template | https://seeto.ai/blog/battle-card-template |
---
## 1. Cognitive Science Principles (Allego)
### Miller's Law (The Magic Number 7 ± 2)
The human brain can hold about 5-9 items in working memory at any given time.
**Implication for wiki:**
- **5-7 top-level categories** is optimal
- Too few → information overload inside each category
- Too many (15+) → overwhelms users
### Hick's Law: More Choices = More Decision Fatigue
The more choices someone has, the longer it takes to decide.
**Implication for wiki:**
- Reduce complexity by guiding to content in fewest possible steps
- **No more than 3 clicks** to find content
- Use filters/search to minimize browsing time
### Chunking Theory
Humans remember and retrieve information better when grouped into meaningful chunks.
**Implication for wiki:**
- Instead of one "Sales Training" folder → separate into Onboarding, Skill Development, Compliance
- Break "Product Information" into product type, industry, use case, competitive landscape
### Recognition vs. Recall
Humans recognize information faster than they recall it from memory.
**Implication for wiki:**
- Use clear, intuitive category names
- "Competitive Battle Cards" not "Docs"
- "Objection Handling Playbooks" not "Resources"
### Hierarchy Depth
**Maximum 3-4 layers** recommended.
Example for Tech:
```
Product Line A → Feature Breakdown → Use Cases → Competitive Comparisons
```
---
## 2. Gong's Built-in Theme Categories
Gong's AI Theme Spotter detects these recurring patterns:
| Theme | Description |
|-------|-------------|
| **Pricing reactions** | Customer responses to pricing discussions |
| **Customer concerns** | Worries and hesitations raised |
| **Customer objections** | Direct pushback on product/process |
| **Strategic business goals** | What customer is trying to achieve |
| **Customer pain** | Problems customer is experiencing |
| **Decision criteria** | Standards used to evaluate options |
### Gong's Multi-Call Analysis Capabilities
- Identify patterns across call sets
- Summarize themes
- Generate insights from large datasets
---
## 3. MEDDPICC Framework
MEDDPICC is a **deal qualification framework**, not a content taxonomy. But its components suggest what knowledge is valuable to capture:
| Letter | Component | Knowledge Implication |
|--------|-----------|----------------------|
| **M** | Metrics | Quantified business outcomes → value props, ROI examples |
| **E** | Economic Buyer | Final authority → stakeholder mapping patterns |
| **D** | Decision Criteria | Evaluation standards → competitive positioning |
| **D** | Decision Process | Internal steps → timeline/process patterns |
| **P** | Paper Process | Procurement/legal → security, compliance handling |
| **I** | Identify Pain | Quantified problems → pain point patterns |
| **C** | Champion | Influencers → champion identification techniques |
| **C** | Competition | Alternatives considered → competitive intelligence |
### MEDDPICC as Wiki Structure?
Could organize wiki by MEDDPICC dimensions, but:
- These are deal-level assessment dimensions, not content categories
- Better to map techniques/patterns TO MEDDPICC stages
- e.g., "This objection handling technique helps identify Economic Buyer"
---
## 4. Sales Playbook Structure (Industry Standard)
Common sections across G2, Salesforce, Pipedrive playbook guides:
### Core Components
1. **Company Overview** - Mission, vision, strategy
2. **Sales Team Roles** - Responsibilities, quotas
3. **Product Offerings** - Solutions, pricing, value props
4. **Sales Methodology** - MEDDPICC, Challenger, SPIN, etc.
5. **Target Customer Profiles** - ICP, buyer personas
6. **Objection Handling** - Common objections + responses
7. **Sales Scripts/Templates** - Call scripts, email templates
8. **Sales Plays** - Situation-specific strategies per stage
9. **Competitive Analysis** - Battlecards, win/loss patterns
10. **Tools, Resources, KPIs** - Tech stack, metrics
### Sales Plays Definition
"Step-by-step strategies for specific situations at each stage of the selling process" - Salesforce
Examples:
- Inbound lead play
- Competitive displacement play
- Expansion/upsell play
- Executive engagement play
---
## 5. Competitive Battlecard Structure (Klue, Seeto)
Industry-standard battlecard structure (one-page format):
| Section | Content |
|---------|---------|
| **Competitor Overview** | Logo, positioning, funding, target market |
| **Product Comparison** | 5-8 key capabilities side-by-side |
| **Where They Win** | 2-3 scenarios where competitor is stronger |
| **Where We Win** | 2-3 scenarios where we're stronger |
| **Win/Loss Zones** | Data-backed winning/losing scenarios |
| **Common Objections** | 2 most frequent + 2-3 sentence responses |
| **What NOT to Say** | 3-5 phrases that damage credibility |
| **One-Line Positioning** | "[Competitor] is [what] for [who], known for [claim]" |
### Key Insight: One Page Constraint
"The constraint of fitting everything on one page forces prioritization of only information reps need in the first five seconds of a competitive objection." - Seeto
---
## 6. Cedar's Existing Extractions
From `apps/server/src/services/field-values/defaults.ts`:
| Field ID | Label | Extraction Type | Metadata Fields |
|----------|-------|-----------------|-----------------|
| `_ii_competitive` | Competitive Positioning | moment | competitorName, outcome (handled/unresolved/deferred) |
| `_ii_pricing` | Pricing Positioning | moment | pricingTechnique, outcome (landed/pushed back/deferred) |
| `_ii_stage_move` | Stage Progression Move | moment | fromStage, toStage |
| `_ii_objections` | Objections & Talk Tracks | moment | objectionTheme, outcome (overcome/deferred/unresolved) |
| `_ii_rep_moments` | Coaching Moments | moment | quality (highlight/watch_out), skillArea |
| `_ii_call_quality` | Call Quality | scorecard | score (1-100) |
### Extraction Descriptions (verbatim from code)
**Competitive Positioning:**
> "When a competitor was mentioned, what exact words did the rep use to respond — and how did the prospect react? Only extract if the rep had a clear counter-move, not just an acknowledgment."
**Pricing Positioning:**
> "What exact words did the rep use when discussing price — and what was the prospect's reaction? Only extract if there was a deliberate framing or anchoring move, not just stating a number."
**Stage Progression Move:**
> "The strategic sales move — a reframe, qualifying question, insight, or commitment ask — that created the conditions for the deal to advance."
**Objections & Talk Tracks:**
> "What exact objection did the prospect raise, and what exact words did the rep use to respond? Include verbatim quotes from both sides."
**Coaching Moments:**
> "A specific moment where the rep demonstrated a replicable sales technique (highlight) or a clear pattern to avoid (watch_out). The quote itself must teach something."
---
## 7. Mapping Cedar Extractions to Wiki Categories
| Cedar Field | Grouping Key | Wiki Category | Example Pages |
|-------------|--------------|---------------|---------------|
| `_ii_competitive` | `competitorName` | `competitors/` | `competitor-x.md`, `competitor-y.md` |
| `_ii_pricing` | `pricingTechnique` | `techniques/pricing/` | `tco-anchoring.md`, `roi-framing.md` |
| `_ii_stage_move` | `toStage` | `stages/` | `discovery.md`, `demo.md` |
| `_ii_objections` | `objectionTheme` | `objections/` | `budget-timing.md`, `security.md` |
| `_ii_rep_moments` (highlights) | `skillArea` | `techniques/` | `discovery-questioning.md` |
| `_ii_rep_moments` (watch_outs) | `skillArea` | `coaching/` (or `techniques/anti-patterns/`) | `common-mistakes.md` |
| `_ii_call_quality` | `userId` | `reps/` (patterns per rep) | `sarah-chen.md` |
---
## 8. Recommended Seeded Categories
Based on research + Cedar extractions.
**Key insight:** Product info, features, ROI, testimonials are ALSO extracted from conversations. Every category can have:
1. **Curated baseline** (official descriptions, marketing content)
2. **Extracted examples** (real moments from calls)
### Top-Level Categories (6)
| Category | Purpose | Curated Baseline | Extracted Examples |
|----------|---------|------------------|-------------------|
| `techniques/` | Sales techniques and plays | Methodology descriptions | `_ii_pricing`, `_ii_rep_moments`, `_ii_stage_move` |
| `objections/` | Objection handling | Framework/playbook | `_ii_objections` |
| `competitors/` | Competitive intelligence | Battlecard template | `_ii_competitive` |
| `stages/` | Stage-specific patterns | Exit criteria, process | `_ii_stage_move` |
| `product/` | Product knowledge | Feature descriptions | Feature mentions, ROI quotes |
| `customers/` | Customer knowledge | Marketing testimonials | Positive moments, use case discussions |
### System Files
| File | Purpose |
|------|---------|
| `_index.md` | Catalog of all pages with summaries |
| `_schema.md` | Wiki conventions and update rules |
| `_proposals/` | Proposed new categories (need approval) |
### Why 6 Categories?
1. **Fits cognitive load guideline** (5-7)
2. **Maps directly to Cedar extractions** (clear backfill path)
3. **Aligns with industry playbook structures** (techniques, objections, competitive intel, stages)
4. **Each category has clear scope** (MECE principle)
---
## 9. Categories NOT Included (Agent Can Propose)
| Category | Reason Deferred |
|----------|-----------------|
| `pain-points/` | Cedar doesn't extract pain points by default; deal-level data better in Turbopuffer |
| `personas/` | Not extracted; could be added to `customers/` or proposed separately |
| `methodology/` | MEDDPICC/Challenger reference; could be added to `product/` or proposed |
| `coaching/` | Could be subcategory of `techniques/` (watch_outs as anti-patterns) |
| `plays/` | Situation-specific strategies; could be subcategory of `stages/` |
| `industry/` | Industry-specific insights; could be proposed if multiple verticals |
| `compliance/` | Approved messaging; could be added to `product/` or proposed |
If these emerge as useful, agent proposes via `_proposals/`.
---
## 10. Category Structure Examples
### `techniques/`
```
techniques/
├── pricing/
│ ├── tco-anchoring.md
│ ├── roi-framing.md
│ └── annual-prepay.md
├── discovery/
│ ├── pain-amplification.md
│ └── question-layering.md
├── stage-moves/
│ ├── commitment-asks.md
│ └── reframes.md
└── multi-threading.md
```
### `competitors/`
```
competitors/
├── competitor-x.md # Full battlecard
├── competitor-y.md
└── status-quo.md # "Do nothing" competitor
```
### `stages/`
```
stages/
├── discovery.md # Patterns, exit criteria, common mistakes
├── demo.md
├── proposal.md
└── negotiation.md
```
---
## 11. Open Questions
1. **Should `techniques/` have subcategories by skill area?** Or keep flat and use frontmatter tags?
2. **How granular should `objections/` be?** By theme (budget, security, timing) or by specific objection?
3. **Should `competitors/` include "status quo" as a competitor?** Industry best practice says yes.
4. **Where do "watch_out" moments go?** Separate `coaching/` category or `techniques/anti-patterns/`?
5. **Should `reps/` be a category?** Per-rep pattern pages vs. keeping rep data only in Turbopuffer.