Effort Estimator

👤 Team Members

Add one or more developers. Each team member has:

• Experience level — Junior, Mid, Senior, or Architect. Higher experience = faster delivery.
• Availability — 10%–100% (default 80%). Models that not all developers are dedicated full-time to a single project.

Multiple developers enable parallelization with diminishing returns per Brooks' Law — communication overhead increases as the team grows.

📅 Start Date

The project start date is used to calculate estimated completion dates. The estimator counts only working days (Monday–Friday) when projecting end dates.

🤖 AI Stack

Select the AI tooling your team will use. The tool models AI impact per development category — boilerplate-heavy work (test classes, data model) sees larger reductions than complex architecture work (integrations, security).

Two sliders fine-tune the AI impact:

• AI Availability — How often the AI is available during work (0–100%)
• Model Quality/Confidence — How reliable the AI output is (0–100%)

Categories

• Data Model — Custom Objects, Fields, Formulas, Rollups, Validation Rules, Record Types, Page Layouts, Junction Objects
• Apex Development — Triggers, Service Classes, Batch/Scheduled/Queueable Apex, Future Methods, REST Endpoints
• LWC / UI — LWC Components, Aura Components, Lightning App Builder Pages, Visualforce Pages
• Flow & Automation — Screen Flows, Record-Triggered Flows, Scheduled/Autolaunched Flows, Flow Orchestration, Approval Processes
• Integration — REST/SOAP/GraphQL, Platform Events, CDC, External Services, MuleSoft, Named Credentials
• User Access & Security — Permission Sets, Sharing Rules, Apex Sharing, FLS, Territory Management
• Experience Cloud — Sites, Guest User Config, Pages, Theming
• OmniStudio — OmniScripts, FlexCards, DataRaptors, Integration Procedures
• AI & Agents — Agentforce Agents/Topics/Actions, Einstein Bots, Copilot Actions, Prompt Templates
• Data Cloud — Streams, DMOs, Segments, Calculated Insights
• Reports & Dashboards — Custom Report Types, Reports, Dashboards

🔢 Count & Complexity

For each component, set the count using the +/− buttons, and the complexity (S / M / C):

• Simple (S) — Straightforward implementation, minimal logic.
• Medium (M) — Standard implementation with moderate logic.
• Complex (C) — Significant business logic, edge cases, or integration points.

Some components (fields, config items, reports) are count-only — complexity is fixed since these are inherently uniform.

⚙️ DevOps & Config

Additional one-time setup activities: SFDX Project Setup, Sandbox Config, CI/CD Pipeline, Unlocked Packages, DevOps Center, and Scratch Org Definitions. A separate config item slider covers Named Credentials, Auth Providers, Connected Apps, Custom Labels, etc.

📏 T-Shirt Size

The estimate maps total raw effort (person-days) to a T-shirt size: XS, S, M, L, XL, or XXL.

📅 Effort Range & PERT

Uses PERT three-point estimation — an industry-standard technique from SAFe and PMI:

• Optimistic — Best-case scenario
• Most Likely — The primary calculated estimate
• Pessimistic — Worst-case (right-skewed, since software tends to overrun more than it underruns)
• PERT Expected — Weighted average: (O + 4M + P) ÷ 6

📆 Estimated Completion Date

Shows a date range based on PERT Expected ± 1 standard deviation (68% confidence interval), counting only working days (Mon–Fri) from your start date. Also shows the single PERT expected date.

📊 Timeline Chart & Breakdown

An SVG chart shows effort breakdown by category, comparing baseline (no AI) vs AI-adjusted timelines. Each category has its own AI impact factor — test classes and data model see the biggest reduction, while integrations and security see less.

🔗 Share & Export

Share Link encodes the full spec in the URL. Download Summary exports a text file. Copy JSON gives the raw spec for API use or re-import.

Tier 1 — No AI / Basic

Manual development or ad-hoc ChatGPT/Gemini/Claude prompts in the browser.

Tier 2 — IDE-Integrated

Einstein for Developers, GitHub Copilot + SF Extensions, Cursor (Agent Mode), Windsurf, Amazon Q Developer. These provide inline completions, chat, and code generation within the IDE.

Tier 3 — Agentic (MCP-Connected)

Cursor + Salesforce MCP, Claude Code + SF MCP, VS Code Copilot Agent + SF MCP. These tools can autonomously deploy, test, and iterate on Salesforce orgs via Model Context Protocol.

Tier 4 — Local / Self-Hosted

Ollama with 7–14B or 32B+ models. Useful for air-gapped environments or cost-conscious teams.

Example JSON

{
  "version": "1",
  "project_name": "Sales Cloud Enhancement",
  "start_date": "2026-05-01",
  "team": [
    { "role": "senior", "availability": 0.9 },
    { "role": "mid", "availability": 0.8 }
  ],
  "ai_stack": "claude_code_mcp",
  "ai_availability": 0.8,
  "model_quality_confidence": 0.85,
  "poc_mode": "ai_assisted",
  "devops": {
    "sfdx_setup": true,
    "cicd_pipeline": true
  },
  "components": {
    "custom_object": { "count": 3, "complexity": "medium" },
    "lwc_component": { "count": 5, "complexity": "complex" },
    "trigger_handler": { "count": 3, "complexity": "medium" },
    "screen_flow": { "count": 4, "complexity": "simple" },
    "rest_outbound": { "count": 2, "complexity": "complex" }
  }
}

Key Fields

Endpoints

Rate Limits

30 requests/minute per IP, 200 requests/minute global. Exceeding returns HTTP 429.

Example: cURL

curl -X POST https://forcerocket.in/api/estimate \
  -H "Content-Type: application/json" \
  -d '{
    "team": [{"role": "senior", "availability": 0.9}],
    "ai_stack": "claude_code_mcp",
    "ai_availability": 0.8,
    "model_quality_confidence": 0.85,
    "components": {
      "lwc_component": {"count": 3, "complexity": "medium"},
      "trigger_handler": {"count": 2, "complexity": "medium"}
    }
  }'

Q: How accurate are these estimates?

The estimates are guidance ranges using PERT three-point estimation — a technique from SAFe and PMI. They are calibrated against real-world Salesforce project data and continuously refined. Use the PERT range to communicate uncertainty.

Q: How does AI impact the estimate differently per category?

AI doesn't speed up all work equally. The tool models this with per-category boost factors. For example, test class generation sees the highest AI acceleration (boilerplate-heavy, pattern-based), while complex integration work or security configuration sees less benefit (requires domain expertise and manual verification).

Q: Does this tool send my data anywhere?

No. All calculations run entirely in your browser. No data is sent to any server. The REST API is a separate service for programmatic access.

Q: Why does my estimate seem too high/low?

Check these common causes:

• AI stack not set — Defaults to the strongest MCP tier. Change to match your actual tooling.
• Complexity mismatch — A "Complex" component takes significantly more time than "Simple". Adjust per component.
• Team level — An all-junior team has a significant penalty. Verify experience levels.
• Availability — Default is 80%. If your team is more/less dedicated, adjust the slider.

Q: What does "PERT Expected" mean?

PERT (Program Evaluation and Review Technique) uses three estimates: optimistic, most likely, and pessimistic. The PERT Expected is a weighted average: (O + 4M + P) ÷ 6, giving more weight to the most-likely scenario. The completion date range uses ±1 standard deviation (68% confidence).

Q: Can I use this for non-Salesforce projects?

No. The component types and categories are specific to the Salesforce ecosystem.