Big 3 Artifacts Quick-Ref
These three diagrams form the backbone of every CTA presentation. Judges spend the most Q&A time probing them. Get them wrong and you fail. Get them right and the remaining artifacts build on a solid base.
1. System Landscape Diagram
Purpose: High-level view of all systems and how they connect. This is your opening move, establishing the full solution scope at a glance.
What to Include
| Element | Details |
|---|---|
| Salesforce products | Sales Cloud, Service Cloud, Experience Cloud, Marketing Cloud, etc. |
| External systems | ERP, data warehouse, marketing tools, DMS, legacy systems |
| Middleware | MuleSoft, cloud services, custom APIs, ESB |
| User channels | Web portals, mobile apps (type: native/hybrid/HTML5), communities |
| Integration arrows | Direction + type (REST, SOAP, ETL, Platform Events, CDC) |
| System ownership | Which team/department owns each system |
| Mobile apps | Include mobile device type and app approach (Salesforce Mobile, custom) |
Layout Rules
| Rule | Why |
|---|---|
| Salesforce at center | It is the solution platform - everything radiates from it |
| Color code consistently | Blue = Salesforce, Gray = external, Orange = integration, Green = channels |
| Label every arrow | Integration type + direction (no unlabeled connections) |
| Single page only | If it needs more detail, create a Level 2 integration diagram separately |
| Include a legend | Explain colors, icons, line styles |
Mandatory Checklist (verify before presenting)
| Check | Pass? |
|---|---|
| Integration layer (MuleSoft/ESB) shown between Salesforce and all external systems? | |
| Legend included with color coding and system status (new/retiring/keeping)? | |
| Every arrow labeled with direction, integration type, and protocol? | |
| All systems from the scenario accounted for? | |
| Salesforce positioned at center? |
Two non-negotiable items
Integration layer and legend are mandatory on every System Landscape. ~95% of CTA scenarios require an explicit middleware layer. A diagram without a legend or with direct point-to-point arrows (no middleware) is a failing artifact pattern.
Common Mistakes
System Landscape gotchas
- Missing integration layer: direct arrows from Salesforce to ERP/DW with no MuleSoft/ESB shown (the most common structural mistake)
- Missing legend: no color key, no system status (new/retiring/keeping), no line style explanation
- Missing systems: forgetting a system mentioned in the scenario (read the scenario twice)
- No integration types: arrows without labels (REST? SOAP? ETL? Platform Events?)
- Missing mobile: the scenario mentions mobile users but you forgot the mobile channel
- No ownership: judges ask “who owns this system?” and you have no answer
- Too detailed: this is Level 1. Do NOT show individual objects or Apex classes here
Judge Q&A Targets
| Question Pattern | How to Prepare |
|---|---|
| ”Why this mix of on-platform vs off-platform?” | Know platform limits that justify external systems |
| ”What if this external system goes down?” | Have a fallback / circuit-breaker answer ready |
| ”Why not consolidate these two systems?” | Articulate the cost/capability trade-off |
2. Data Model / ERD
Purpose: Logical data model showing standard objects, custom objects, relationships, and cardinality. Judges scrutinize this heavily.
What to Include
| Element | Details |
|---|---|
| Standard objects | Account, Contact, Opportunity, Case, Lead, etc. |
| Custom objects | Clear naming + stated purpose |
| Relationships | Master-Detail vs Lookup with cardinality (1:1, 1:N, N:N via junction) |
| Record ownership | Who owns records of each object type |
| OWD settings | Private, Public Read Only, or Public Read/Write per key object |
| Key fields | External IDs, formula fields, roll-up summaries (only architecture-driving fields) |
| Data volumes | Approximate record counts (flag LDV objects with 1M+ records) |
| External objects | If using Salesforce Connect |
Notation Reference
Layout Rules
| Rule | Why |
|---|---|
| Crow’s foot notation | Standard ERD cardinality that judges expect |
| Group related objects | Sales objects together, Service objects together |
| Annotate OWD on diagram | Or in a companion Google Sheets table if too busy |
| Highlight junction objects | Many-to-many relationships need explicit junction objects |
| Flag LDV objects visually | Star/icon marker on objects exceeding 1M records |
| Show ownership arrows | Critical for judges to evaluate sharing model |
Master-Detail vs Lookup Decision
| Factor | Master-Detail | Lookup |
|---|---|---|
| Cascade delete | Yes - child deleted with parent | No - child orphaned |
| Sharing inheritance | Child inherits parent sharing | Independent sharing |
| Roll-up summaries | Supported | Not supported (use Flow/Apex) |
| Required relationship | Always required | Can be optional |
| Reparenting | Not by default (can enable) | Always allowed |
| Lock contention risk | Higher on high-traffic parents | Lower |
| CTA rule of thumb | Use when child cannot exist without parent | Use for loose associations |
Common Mistakes
Data Model gotchas
- No cardinality: showing relationships without 1:N or M:N notation
- Missing OWD: the ERD exists but does not show org-wide defaults
- Ignoring LDV: objects with millions of records shown with no volume annotation
- Over-using Master-Detail: creates lock contention on high-traffic parent objects
- No external IDs: migration and integration depend on these; judges will ask
- Wrong level of detail: showing every field vs showing only architecture-driving fields
Judge Q&A Targets
| Question Pattern | How to Prepare |
|---|---|
| ”Why Master-Detail vs Lookup here?” | Know the 6 comparison factors above |
| ”How do you handle 10M+ records on this object?” | Indexing, skinny tables, archiving, Big Objects |
| ”Walk me through the migration load order” | Parent objects first, junction objects last |
| ”What is your query optimization for LDV?” | Selective queries, custom indexes, async processing |
3. Role Hierarchy + Sharing Model
Purpose: The organizational hierarchy driving record-level visibility, combined with the full sharing architecture. This is the #1 failure domain.
What to Include
| Element | Details |
|---|---|
| Role hierarchy tree | CEO down to portal users, as a top-down tree |
| OWD summary | Per-object defaults (Private, Public Read Only, Public Read/Write) |
| Sharing rules | Criteria-based and ownership-based, with which objects |
| Sharing mechanisms | Apex Managed Sharing, Manual Sharing, Team Sharing |
| Territory Management | If scenario involves geographic or named-account territories |
| Permission Sets | Key PSGs for feature access |
| Portal/Community roles | Where external users sit in the hierarchy |
| Profile-to-role mapping | Which profiles are assigned to which roles |
Sharing Model Layers (Bottom-Up)
The golden rule
Start restrictive (Private OWD), then open access with sharing rules. Never start with Public Read/Write and try to restrict down - Salesforce does not support “closing” access below OWD.
OWD Decision Quick-Ref
| If… | Set OWD to… | Then open with… |
|---|---|---|
| Most users should NOT see most records | Private | Sharing rules, role hierarchy, teams |
| Most users need to read but not edit | Public Read Only | Sharing rules for edit access |
| Everyone can see and edit everything | Public Read/Write | Nothing (already fully open) |
| Child object follows parent visibility | Controlled by Parent | Parent OWD + sharing rules |
Common Mistakes
Sharing model failures - the #1 killer
- No OWD justification: saying “Private” without explaining why
- Forgetting portal users: 50K community users in Private OWD = sharing explosion
- Flat hierarchy: too few roles, not reflecting actual data access needs
- Org-chart copy: role hierarchy should reflect data access, NOT reporting structure
- Ignoring scale: sharing rules that work for 100 users break at 50,000 users
- No sharing rules documented: judges expect you to name specific rules, not just say “sharing rules”
- Missing Apex Managed Sharing: when standard sharing mechanisms are insufficient
Judge Q&A Targets
| Question Pattern | How to Prepare |
|---|---|
| ”Walk me through how user X sees record Y” | Trace the path: OWD -> role -> sharing rule -> teams |
| ”What if you have 50K portal users?” | Sharing sets, high-volume portal users, sharing group limits |
| ”Why Private instead of Public Read Only?” | Business justification for restriction level |
| ”What happens to sharing when a user changes roles?” | Sharing recalculation, implicit vs explicit shares |
| ”How does Territory Management fit?” | When geography-based or named-account access is needed |
Big 3 Interaction Map
These three diagrams are tightly connected. Judges will test whether your artifacts stay consistent with each other.
| Cross-reference Check | What to Verify |
|---|---|
| System Landscape <-> Data Model | Every external system in the landscape has data that flows to/from objects in the ERD |
| Data Model <-> Role Hierarchy | Every object with Private OWD has a sharing mechanism defined in the role hierarchy artifact |
| Role Hierarchy <-> System Landscape | Portal users in the role hierarchy connect to the Experience Cloud portal in the landscape |
Sources
- Salesforce Diagramming Framework - architect.salesforce.com
- Salesforce Role Hierarchy Reference - architect.salesforce.com
- Salesforce Data Model Notation - developer.salesforce.com
- CTA Diagramming Framework - FlowRepublic
- CTA Certification Guide & Tips - Salesforce Ben
- Brief Insights from a CTA Board Judge - Dinesh Yadav
- Deep Dive into JourneyToCTA - Apex Hours
Personal study notes for the Salesforce CTA exam. Content compiled from VJ's study notes, official Salesforce documentation, community sources, and online publicly available content, then organized and presented with AI assistance. Not affiliated with Salesforce. © 2025–2026 VJ Srivastava.