“We automated 50 processes last year, but nobody uses them.”

This is the most painful sentence a Head of Automation can hear. It usually happens when you measure success by quantity (number of bots) rather than value (hours saved, accuracy improved).

To scale RPA successfully, you need a Center of Excellence (CoE)—a governing body that decides what to automate, how to measure it, and who is responsible.

---

The Trap: Automating for the Sake of Automation

Not every painful process is a good candidate for RPA.

The Bad Candidate Checklist ❌

• The Process Changes Monthly: Creating a bot for a moving target is a maintenance nightmare.
• Low Volume, High Complexity: Spending 3 months to automate a task that takes 1 hour a month.
• Requires “Human Intuition”: “Just look at the email and decide if the customer owns seems angry.” Bots can’t do feelings (yet).
• Broken Process: Automating a bad process just makes a bad process faster. Optimize first, automate second.

---

The Solution: The Automation Suitability Matrix

How do you scientifically decide if a process is worth identifying? Use this formula.

$$\text{Suitability} = \frac{\text{Volume} \times \text{Stability} \times \text{Rule-Based}}{\text{Complexity}}$$

Where (each scored 1–5):

• Volume — Transactions per month
• Stability — How often the process changes
• Rule-Based — How deterministic the logic is
• Complexity — Technical difficulty to automate

Scoring Guide

Metric	1 (Low)	5 (High/Best)
Volume	< 50 / month	> 10,000 / month
Stability	Changes weekly	Hasn’t changed in 2 years
Rule-Based	”It depends on the vibe"	"If A then B, always”
Complexity	10 apps, Citrix, Handwriting	Excel + Web, Structured Data

Decision Rule:

• Score > 15: Green Light. Automate immediately.
• Score 5-15: Yellow Light. Optimize first or wait for higher volume.
• Score < 5: Red Light. Do not automate.

---

Building an RPA Center of Excellence (CoE)

A CoE isn’t just a bureaucracy; it’s the engine room. It typically consists of these key roles:

1. The Core Roles

Role	Responsibility
RPA Sponsor (Head)	Holds the budget, defines strategy, evangelizes value to the C-suite.
Business Analyst (BA)	The bridge. Translates “business speak” into PDDs (Process Definition Docs). Calculates ROI.
Solution Architect (SA)	The tech lead. Designs the SDD, ensures scalability, code reviews.
RPA Developer	Builds the bots based on SDD.
Support / Controller	Monitors the bots in production, handles exceptions (Hypercare).

2. The Interaction Model

sequenceDiagram
    participant BU as Business Unit
    participant BA as Business Analyst
    participant SA as Solution Architect
    participant Dev as Developer

    Note over BU: Identifies Pain Point
    BU->>BA: Submit Idea
    
    activate BA
    Note over BA: Check Suitability
    BA->>BA: Feasibility Assessment
    BA->>BA: ROI Calculation
    BA->>SA: Handover for Tech Review
    deactivate BA

    activate SA
    Note over SA: Check Architecture
    SA->>SA: Technical Feasibility
    SA->>SA: Create Implementation Plan
    SA->>Dev: Assign Task
    deactivate SA

    activate Dev
    Dev->>Dev: Build & Test
    Note over Dev: Solution Ready
    deactivate Dev

---

Calculating ROI: The Math of Automation

ROI isn’t just “we fired 2 people.” (In fact, RPA rarely leads to direct firing; it leads to capacity creation).

The Hard ROI Formula

$$ROI = \frac{\text{Cost Savings} - \text{Total Cost of Ownership (TCO)}}{\text{TCO}} \times 100$$

1. Cost Savings

• Hours Returned: (Time per transaction × Volume) × Hourly Rate
• Error Reduction: Cost of fixing one error × Error rate × Volume
• SLA Penalty Avoidance: Fines saved by processing faster

2. Total Cost of Ownership (TCO)

• License Costs: UiPath/Automation Anywhere/BluePrism licenses
• Infrastructure: Servers, SQL Databases, VMs
• Development Costs: Developer hours × Rate
• Maintenance: Typically 20-30% of development cost per year

Example Calculation

Process: Invoice Processing

• Volume: 2,000 invoices/month
• Time: 10 mins/invoice
• Rate: $30/hour
• Current Cost: (2000 × 10/60) × $30 =$10,000 / month = $120,000 / year

Automation Cost:

• License: $8,000 / year
• Dev: $20,000 (one time)
• Maintenance: $5,000 / year
• Total Year 1: $33,000

Year 1 ROI:

• Savings: $120,000 (assuming 100% automation)
• Cost: $33,000
• Net Benefit: $87,000 (263% ROI)

---

Governance: Stopping “Bot Sprawl”

Without governance, you get “Shadow IT”—random scripts running on people’s desktops that break when a password changes.

CoE Best Practices for Governance

1. Centralized Repository: All code must be in Git/SVN. No .xaml files on desktops.
2. Code Reviews: No bot goes to production without a review by an Architect.
3. Strict Access Control: Bots have their own Active Directory accounts (e.g., SVC_BotFinance), not user accounts.
4. Change Management: When SAP updates, the CoE must know before the update, not after the bots crash.

---

Conclusion

RPA is not a magic wand; it’s a force multiplier.

• Without a CoE: You have a collection of scripts and a headache.
• With a CoE: You have a digital workforce that scales, complies with security policies, and delivers measurable value.

Start with the Matrix. Score your ideas. Build only what matters.