Scope Analysis: The First Step in AI Adoption for Any Business

# Scope Analysis of Work Processes: The First Step in AI Adoption for Any Role

## Why Everyone Skips This Step

Here's how AI adoption usually starts in a company: leadership reads about Claude Code and Codex, buys subscriptions, holds a hackathon, the team starts "trying" the tools. After 3 months — a report: "We tried it, results exceeded expectations, but there are no concrete metrics."

This is a failure. Because **AI adoption starts not with tools, but with process analysis**. And this analysis is done by the people who perform these processes every day. People of any role: developers, marketers, salespeople, HR, finance, lawyers, CEOs.

Without this analysis, you don't know **what to automate**. And you start automating blindly.

---

## What Is a Scope Analysis

A scope analysis is **building a complete map of current manual processes** before you start thinking about AI. It answers simple questions:

1. What exactly do you do every day? Step by step.

2. Which tasks are repetitive?

3. Where do you lose the most time?

4. Which of this is routine (automatable)?

5. Which of this is judgment (requires your judgment)?

6. What are the "before" metrics (time, iterations, rework)?

Without answers to these questions, you don't know what to automate. You throw AI at the first task that comes along and get a wow effect at the hackathon, but zero effect in real work.

---

## The Practical Process: Journal → AI Auditor → Loop

Here's a simple process that works for **any role**. You don't need a team of analysts. You don't need weeks of work. This is a semi-automated process that you do yourself with AI.

### Step 1: Work Day Journal (30 minutes)

Open a document and write **in maximum detail** what you did today. Not how it should be according to processes. But how it actually was.

**Example for a marketer:**

```

09:00 - Opened email, 15 minutes sorting through it

09:15 - Went into analytics, checked campaign metrics

09:45 - Opened Excel, copied data from analytics

10:00 - 1.5 hours making a report for management

11:30 - Team call, discussed new campaign

12:00 - Lunch

13:00 - Wrote social media posts, 3 posts in 2 hours

15:00 - Launched A/B test, configured parameters

15:45 - Responded to comments on social media, 1 hour

16:45 - Filled in CRM for leads, 45 minutes

17:30 - Left

```

**Example for a salesperson:**

```

09:00 - Checked email, replied to 10 emails

09:30 - Opened CRM, checked tasks for the day

10:00 - 5 cold lead calls, 2 hours

12:00 - Lunch

13:00 - Prepared proposal for client, 1.5 hours

14:30 - Client call, product demo

15:30 - Follow-up after call, proposal email

16:00 - Filled in CRM with call results

16:45 - Sales funnel analysis, 45 minutes

17:30 - Left

```

**Example for a developer:**

```

09:00 - Checked email and Slack

09:30 - Standup, 15 minutes

09:45 - Opened Jira task, read requirements, 30 minutes

10:15 - Studied related code, 1 hour

11:15 - Wrote code, 2 hours

13:15 - Lunch

14:00 - Wrote tests, 1 hour

15:00 - Local testing, 30 minutes

15:30 - Created PR, 10 minutes

15:40 - Waited for review, 2 hours

17:40 - Review fixes, 30 minutes

18:10 - Left

```

Don't filter. Don't edit. Write as is. Include the small stuff: "15 minutes sorting email", "45 minutes filling in CRM". The routine is hidden precisely in the small details.

### Step 2: Feed the Journal to the AI Auditor (5 minutes)

Take your journal and feed it to AI with this prompt:

```

You are a work process auditor. I've given you a description of my work day.

Your task:

1. Identify repetitive tasks (routine)

2. Identify tasks requiring my judgment (judgment)

3. Find tasks that take a lot of time

4. Find tasks that can be automated

5. Ask me 10-15 clarifying questions to better understand my workflow

Response format:

- Task classification (routine/judgment/mixed)

- List of potential automations

- Clarifying questions

```

**What you'll get:**

AI will analyze your journal and output:

- Task classification (what's routine, what's judgment)

- List of potential automations

- **Clarifying questions** (this is the most important part)

**Example clarifying questions for a marketer:**

```

1. The management report — is this a weekly task? Which template?

2. Social media posts — based on what brief? Is there a style/tone?

3. A/B tests — which parameters do you configure? Based on what criteria?

4. Comment responses — is there an FAQ or template answers?

5. CRM filling — which fields? Where do you get the data?

6. Metrics analysis — which metrics do you look at? What are you looking for?

7. Team calls — are these regular meetings? What frequency?

```

### Step 3: Answer the Questions (20 minutes)

Answer AI's questions. Briefly. In bullet points. Don't write essays. 1-2 sentences per question.

**Example answers:**

```

1. Yes, weekly. Template: campaign metrics + conclusions + plan for the week.

2. Brief from management. Tone: friendly but expert.

3. I configure: headlines, images, CTA. Criterion: conversion.

4. FAQ exists but is outdated. I write situationally.

5. Fields: name, company, source, status. Data from email and calls.

6. I look at: CTR, conversion, lead cost. I look for anomalies.

7. Yes, weekly. Plus ad-hoc as needed.

```

### Step 4: Repeat the Loop (2-3 iterations)

Feed the answers back to AI:

```

Here are my answers to your questions. [answers]

Now:

1. Update the task classification with the new data

2. Suggest specific automations for each routine task

3. For each automation, suggest "before" and "after" metrics

4. If more clarification is needed — ask more questions

```

**Repeat 2-3 times** until AI produces a sensible picture.

**Example result after 3 iterations:**

```

TASK CLASSIFICATION:

Routine (automatable):

- Weekly report preparation (1.5 hours/week)

- Writing posts from brief (2 hours/day)

- CRM filling (45 min/day)

- Standard comment responses (30 min/day)

Judgment (stays with you):

- Metrics analysis and anomaly detection

- Strategic campaign decisions

- Creative content decisions

- Team negotiations

Mixed:

- A/B test configuration (routine — setup, judgment — hypothesis selection)

- Funnel analysis (routine — data collection, judgment — conclusions)

SUGGESTED AUTOMATIONS:

1. Weekly report

   - What: AI pulls data from analytics, generates report from template

   - "Before" metric: 1.5 hours/week

   - "After" metric: 10 minutes for verification

   - Expected savings: 5 hours/month

2. Post writing

   - What: AI generates drafts from brief, you verify

   - "Before" metric: 2 hours/day (6 posts)

   - "After" metric: 30 minutes for verification

   - Expected savings: 35 hours/month

3. CRM filling

   - What: AI parses email and calls, auto-fills fields

   - "Before" metric: 45 min/day

   - "After" metric: 5 minutes for review

   - Expected savings: 15 hours/month

TOTAL: ~55 hours saved per month

```

### Step 5: Measure "Before" Metrics (15 minutes)

For each task you plan to automate, measure:

- **Execution time** (how much you actually spend)

- **Number of iterations** (how many times you redo)

- **Rework percentage** (how many tasks come back for revision)

This is the **baseline**. Without it, you can't prove ROI.

**Example:**

```

Task: Weekly report

- Time: 1.5 hours

- Iterations: 1-2 (management requests edits)

- Rework: 30% (comes back for revision)

Task: Writing posts

- Time: 20 minutes per post × 6 posts = 2 hours

- Iterations: 1-3 (edits from management)

- Rework: 40%

Task: CRM filling

- Time: 45 minutes

- Iterations: 1

- Rework: 10% (data errors)

```

---

## What to Do with the Result

After scope analysis you have:

**1. Process map.** What you do, how much time, what's routine, what's judgment.

**2. Automation list.** Specific tasks that can be automated, with priorities.

**3. "Before" metrics.** Baseline for comparison.

**4. Expected ROI.** How many hours you'll save.

Now you're ready for the next step: **AI implementation for specific tasks**. Not "trying Claude Code". But **automating specific routine** with specific success metrics.

---

## Examples for Different Roles

### Marketer

**Routine:** report preparation, content generation from brief, CRM filling, standard comment responses, A/B test setup from template.

**Judgment:** strategic campaign decisions, metrics anomaly analysis, creative decisions, test hypothesis selection.

**Expected savings:** 50-60 hours/month.

### Salesperson

**Routine:** proposal preparation from template, follow-ups, CRM filling, funnel analysis (data collection), client research.

**Judgment:** negotiations, strategy selection for specific clients, understanding client pain points, deciding when to escalate.

**Expected savings:** 40-50 hours/month.

### HR

**Routine:** resume screening by criteria, offer preparation from template, onboarding docs, hiring reports, standard candidate question responses.

**Judgment:** conducting interviews, cultural fit assessment, salary negotiations, strategic hiring decisions.

**Expected savings:** 30-40 hours/month.

### Developer

**Routine:** coding from specification, writing tests, running CI/CD, debugging from logs, filling in Jira.

**Judgment:** architectural decisions, technology selection, verifying solution correctness, understanding business requirements.

**Expected savings:** 60-80 hours/month.

### CEO

**Routine:** board report preparation, metrics analysis, competitor intelligence gathering, investor email drafts, presentation preparation.

**Judgment:** strategic decisions, negotiations, team management, market understanding.

**Expected savings:** 20-30 hours/month.

### Lawyer

**Routine:** contract review by checklist, standard document preparation, case law analysis by template, registry filling.

**Judgment:** law interpretation for specific cases, defense strategy, negotiations, risk assessment.

**Expected savings:** 40-50 hours/month.

---

## Why This Works for Any Role

The process is universal because:

**1. All roles have routine.** It doesn't matter if you're a marketer, salesperson, developer, or CEO. Everyone has repetitive tasks that can be described algorithmically.

**2. All roles have judgment.** Everyone has tasks requiring judgment, experience, and intuition. This doesn't get automated. This stays with you.

**3. AI is good at classification.** Feed it a journal — AI identifies routine and judgment. This is a basic LLM task.

**4. The clarification loop works.** AI asks questions, you answer, AI refines. After 2-3 iterations you get a sensible picture.

**5. Metrics are universal.** Time, iterations, rework — this works for any role.

---

## How Long It Takes

**Full scope analysis cycle:** 2-3 hours.

- Journal: 30 minutes

- First iteration with AI: 25 minutes (5 minutes prompt + 20 minutes answers)

- Second iteration: 25 minutes

- Third iteration: 25 minutes

- Measuring "before" metrics: 15 minutes

- Finalizing the result: 20 minutes

**Total: ~2.5 hours.** This isn't weeks of work. It's one evening or morning.

---

## What to Do If You Manage a Team

If you're a team lead, department head, or CEO — you do scope analysis **for each team member**.

**Process:**

1. Each team member writes their own journal (30 minutes)

2. Each feeds it to the AI auditor and goes through the clarification loop (2 hours)

3. You collect results from everyone

4. You analyze: where are the common routine patterns, where are unique tasks

5. You prioritize automations: what to automate first for the whole team, what individually

**Result:** process map for the entire team, automation list with priorities, "before" metrics for each person.

**Time:** 1 day for a team of 5-10 people.

---

## Why This Is the First and Mandatory Step

Without scope analysis you:

- Automate the wrong things

- Add AI to the old process instead of restructuring

- Can't prove ROI

- Get team resistance

- Fail the implementation within 6 months

With scope analysis you:

- Know what to automate

- Restructure the process for AI

- Have baseline metrics for comparison

- Get team buy-in (they see for themselves how much routine they do)

- Implement systematically, not chaotically

---

## Conclusion

Scope analysis is **the first and mandatory step** in AI adoption. It works for **any role**: marketer, salesperson, developer, HR, CEO, lawyer.

The process is simple and semi-automated:

1. Write a work day journal (30 minutes)

2. Feed it to the AI auditor (5 minutes)

3. Answer clarifying questions (20 minutes)

4. Repeat the loop 2-3 times (1 hour)

5. Measure "before" metrics (15 minutes)

**Total: 2-3 hours.** And you have a process map, automation list, metrics for comparison, and expected ROI.

Without this step, you implement AI blindly. With this step, you implement AI systematically and can prove results with numbers.

This isn't bureaucracy. This is **2-3 hours of work that determine the success or failure of the entire project**.