5. Step-by-step and chain prompting

**Why big tasks are hard for a single prompt**

Imagine asking a friend to plan your entire wedding in one sentence. They would either be overwhelmed or give you a vague, generic answer. AI has the same challenge.

When you send one long, complicated prompt, the AI tries to do everything at once. It may skip steps, forget constraints, or produce a shallow result. Breaking the work into clear, ordered steps — just like a recipe — gives the AI a manageable path to follow.

Consider Abebe, who owns a small electronics repair shop in Merkato. He wants the AI to help him write a full business plan. If he says *"Write me a complete business plan for my shop"*, the result will likely be a generic template. But if he guides the AI step by step — first the market analysis, then the pricing, then the marketing plan — each section will be thoughtful and specific to his situation.

**How step-by-step prompting works**

Step-by-step prompting means you explicitly ask the AI to work through a problem in a defined order. There are two common approaches:

**Approach A — You list the steps yourself:**

You tell the AI exactly what to do at each stage. This gives you full control.

*Example:*

*"Help me prepare for my job interview at a bank. Do it in three steps:

1. List five common interview questions for a bank teller role.

2. Write a strong, two-sentence answer for each question.

3. Give me two tips on how to appear confident during the interview."*

**Approach B — You ask the AI to think step by step before answering:**

You add the phrase *"Think step by step"* or *"Work through this step by step before giving your final answer."* This tells the AI to reason carefully before replying, which reduces errors on complex tasks.

*Example:*

*"A telebirr agent in Addis Ababa charges 1% commission on every transfer. If a customer sends 4,500 birr and another sends 12,000 birr, what is the total commission? Think step by step."*

Using this phrase alone can dramatically improve accuracy on maths and logic problems.

**What is chain prompting?**

Chain prompting means using several separate prompts in a sequence, where the output of the first prompt becomes the input for the second, and so on — like a relay race.

This is different from step-by-step prompting, where you give one prompt with multiple steps. In chain prompting, you send multiple prompts, one at a time, building on what the AI just produced.

Here is a real example. Sara wants to launch an online Amharic tutoring service in Addis Ababa. She uses a chain of three prompts:

**Prompt 1 (Research):**

*"List five common challenges Ethiopian students face when learning English. Be specific."*

→ The AI produces a list of five challenges.

**Prompt 2 (Plan) — uses the output of Prompt 1:**

*"Based on these five challenges [paste the list], write a 3-month lesson plan for beginner English students. Focus on solving challenge numbers 1, 2, and 4."*

→ The AI produces a focused lesson plan.

**Prompt 3 (Marketing) — uses the output of Prompt 2:**

*"Using the lesson plan above [paste the plan], write a short promotional post in Amharic for my WhatsApp business group that explains what students will learn and how to register via telebirr."*

→ The AI produces a ready-to-send marketing message.

Each prompt is simple on its own, yet the chain produces a rich, connected end result that no single prompt could achieve.

**Building a reusable chain template**

Once you find a chain that works well for a recurring task, you can turn it into a reusable template. Just save the sequence of prompts and replace the variable parts each time.

Here is a simple 3-prompt chain template for writing a product launch announcement — useful for any small shop owner in Ethiopia:

---

**PROMPT 1 — Research**

*"I sell [product] at my shop in [city]. List three reasons why a customer in [city] would want to buy this product. Be specific to Ethiopian customers."*

**PROMPT 2 — Draft announcement**

*"Using these three reasons [paste reasons], write a short launch announcement (under 80 words) in Amharic for my [product]. Mention that customers can pay with telebirr."*

**PROMPT 3 — Final polish**

*"Read this announcement [paste announcement]. Make it more exciting. Add one sentence that creates urgency — for example, a limited-time offer. Keep it under 90 words."*

---

This three-step chain takes a simple product idea all the way to a finished, exciting, customer-ready message. Almaz used this exact chain to write the launch announcement for her new tilet (sesame) cooking oil — and got 14 new orders in one day via telebirr.