Last updated: 10 November 2026
The best way for a small sales team to use AI for lead generation is to start with the customers you already have. Ask an AI assistant what your best customers have in common, turn that into an ideal customer profile, build a target list from sources you are allowed to use, and score each company against the profile. AI speeds up every step. You decide who goes on the list and whether you may contact them.
Each step comes with a prompt you can paste into ChatGPT, Claude, Google Gemini or Microsoft Copilot; replace the words in square brackets with your own details. The examples come from an invented company that sells inventory software to small retailers. When your list is ready, read our PDPA and cold outreach checklist before you contact anyone, and see our guide to AI cold email prompts for the next step. For the whole sales process, start with our practical guide to AI for sales.
What is lead generation, and where does AI help?
Lead generation is the work of finding people or companies that may want what you sell and starting a conversation with them. In a small team it usually means research, list building and first outreach, done between customer meetings.
AI helps most with the research. It can find the pattern in your best customers, draft a profile from it, summarise what each prospect does and suggest a reason to get in touch. It cannot tell you who you are allowed to contact, and it should not be the source of anyone’s email address or phone number.
Step 1: How do you find what your best customers have in common?
Give the assistant a short description of your best customers and ask it to find the patterns. Use company-level facts only, such as industry, size, what they bought, how long the sale took and why they chose you. Leave out the names and contact details of individual people.
Pick ten to twenty customers who bought, stayed and would buy again. If you can, add a few who did not work out, so the assistant can see the difference.
You are a B2B sales analyst. Below are short descriptions of our best customers and of a few customers who did not work out. Find what the best customers have in common: industry, size, the situation when they bought, the problem they had and who made the decision. Then list what the poor-fit customers had in common. Mark any pattern that rests on fewer than three examples as weak. Do not invent details.
[Paste customer descriptions here]
Check the answer against your CRM by hand. A pattern that holds for most of your best customers is useful. A pattern that appears in a single account is a coincidence.
Step 2: How do you write the ideal customer profile?
An ideal customer profile is a one-page description of the companies most likely to buy from you and keep buying. It describes the company. A buyer persona describes the person inside it, and you can add that later.
Ask the assistant to turn the patterns from step 1 into a profile with the parts below.
| Part | Question it answers | Invented example |
|---|---|---|
| Company | What industry, size and location fit? | Retailers with two to ten outlets in Singapore |
| Situation | What is happening when they need us? | Stock counts done by hand in spreadsheets |
| Problem | What does it cost them? | Out-of-stock items and slow month-end counts |
| Buyer | Who decides and who uses it? | The owner or the operations manager |
| Proof | Which of our results apply? | A result we can show for a similar retailer |
| Disqualifiers | Who should we not chase? | Single-outlet shops, and chains with their own IT team |
Do not skip the disqualifiers. A profile that says who not to chase saves more time than one that only says who to chase.
Using the patterns above, write an ideal customer profile for [my company] on one page with these parts: company, situation, problem, buyer, proof and disqualifiers. Keep each part to two sentences. Where the evidence is thin, say so.
Share the profile with the team and ask what is missing. People who talk to customers every day will spot gaps quickly.
Step 3: Where can you find companies that match?
Start with sources you already hold, then add public ones. Your own CRM and past enquiries often contain companies that fit but were never followed up. Existing customers can introduce similar businesses. Public company websites, industry association directories and news about openings or expansions show where a need may be starting.
Two rules keep this step safe:
- Check each source’s terms of use. Some websites and directories do not allow scraping or bulk copying.
- If you buy or receive a list, ask where the data came from and whether the people on it agreed to be contacted for your purpose.
Let AI organise the contact details you have found yourself. An assistant that is asked for an email address may produce one that looks right and belongs to nobody, or to the wrong person. Give it the company names and websites that you collected and ask it to tidy the list, remove duplicates and group the companies by how closely they fit the profile.
Step 4: How do you research and score the list?
Paste the text from each company’s website into the assistant and ask it to compare the company with your profile. Use a simple scale, ask for evidence behind each score and ask it to mark anything it is guessing.
You are a B2B sales researcher. Here is our ideal customer profile: [paste profile]. Below is text from a company’s website. Score the company from 0 to 2 on each of company, situation, problem and buyer, and give one sentence of evidence from the text for each score. Then suggest one specific reason we might get in touch. If the text does not say, write ‘unknown’ and do not guess.
[Paste website text here]
A scored list for the invented retailer example might look like this.
| Company (invented) | Evidence from the website | Score out of 8 | Next step |
|---|---|---|---|
| Retailer A | Five outlets, an online store and stock transfers between outlets | 7 | Priority: draft a first email |
| Retailer B | One outlet and no online store | 2 | Skip: single outlet |
| Retailer C | Three outlets and no information about stock | 4 | Research further before contacting |
Use the score to sort the list. Read the top companies yourself, since the assistant only knows what the website says, and a company with a thin website may score low even though it fits.
Step 5: How do you check who you may contact?
Check permission before you write any message. Researching a company from its public website is a different step from sending someone a marketing message, and the rules apply at the second step.
In Singapore, marketing messages sent to Singapore telephone numbers fall under the Do Not Call provisions of the PDPA, which means checking the register or holding clear consent. The PDPC guidelines exclude a message sent to an organisation for its own purposes, but that exclusion does not cover a message that offers something for the person’s personal use. Bulk unsolicited commercial email and text messages are also covered by the Spam Control Act, and the PDPA applies to the personal data you collect and use. Our PDPA and cold outreach checklist sets out the checks in order. This is general information and not legal advice.
Step 6: How do you keep the list useful?
Review the list every month. Record which companies replied, which became customers and which dropped out, then give those results to the assistant and ask what the profile should change. A profile built from your first ten customers will drift as you win more.
Here is our current ideal customer profile and a table of this month’s outcomes: companies contacted, replies, meetings and sales, with the profile score for each. Which parts of the profile predicted a reply or a sale, and which did not? Suggest changes to the profile and say how confident you are, given the small numbers.
Small teams have small numbers, and five replies do not prove a pattern. Wait for changes that show up over several months before you rewrite the profile.
What are the limits of AI lead generation?
- It can invent contact details, company facts and news. Check anything you will use against the company’s own site.
- It works from the text you provide, so its view of a company is only as good as the page you pasted.
- It does not know your market the way your salespeople do. Ask them to review the profile and the top of the list.
- It cannot tell you whether you may contact someone. That check is yours.
- Website text can be out of date. Look for a recent update before you rely on a trigger such as an expansion.
Frequently asked questions
What is an ideal customer profile?
An ideal customer profile is a one-page description of the companies most likely to buy from you and keep buying. It covers their size, industry, situation, problem and decision maker, and it lists the companies you should not pursue. A buyer persona describes the person who decides, and the profile describes the company.
Can AI find leads for me?
AI can help you research and rank companies, summarise what they do and suggest a reason to get in touch. It should not be your source for names, email addresses or phone numbers, because an assistant can produce details that look real and are wrong. Find contacts from sources you can verify and are allowed to use.
Is AI lead generation allowed under the PDPA in Singapore?
Researching a company from its public website is a different step from contacting someone. The PDPA and the Do Not Call provisions apply when you collect personal data or send marketing messages, so check the source of each contact and the rules for the channel before you send anything. Our PDPA and cold outreach checklist covers the steps. This is general information and not legal advice.
How long should the first target list be?
Make it a size your team can work through properly in a couple of weeks. A short list that you research and contact well tells you more than a long list you never finish, and you can add companies once you see what gets replies.
Do I need paid tools to do AI lead generation?
No. An AI assistant, your CRM or a spreadsheet and the company websites are enough to start. Paid data tools can help later, but they do not remove the need for a clear profile or for checking who you may contact.
Where does this guide fit in our AI for sales series?
Lead generation is the first stage of selling with AI: deciding who to approach and building a list worth contacting. Start with the practical guide to AI for sales, use the PDPA and cold outreach checklist and the AI cold email prompts to contact your list, and plan the messages that come after with our follow-up cadence guide. We also practise this flow in our two-day Mastering AI for Sales Growth and Team Dynamics course.
Build your AI lead generation flow in class
QD Academy’s two-day Mastering AI for Sales Growth and Team Dynamics course covers AI-powered qualified lead generation and automated outreach, taught by practitioners and funded for eligible learners. Check the course page for the latest dates, fees and funding.