AARRR Funnel: Measure the 5 Growth Stages of Your Business, with a Prompt Guideline So AI Can Find the Leaks
You are already doing marketing, but is the business really growing? AARRR splits the customer journey into 5 stages, from first hearing about you to generating revenue, so you can see where customers drop off. This article shows what data to give AI and includes prompts you can copy into ChatGPT, Gemini or Claude right away.
What Is AARRR?
AARRR, nicknamed Pirate Metrics because it sounds like a pirate's cry, is a growth framework presented by Dave McClure, investor and founder of 500 Startups, in a five-minute talk called Startup Metrics for Pirates at Ignite Seattle in 2007. It splits the customer journey into 5 stages: Acquisition, Activation, Retention, Referral and Revenue. It was designed for tech startups, but shops and service businesses can adapt the definition of each stage. Its strength is that it makes you look at numbers tied to sales, not just views or likes.
What the AARRR Funnel Looks Like
Many people enter at the top of the funnel and drop off gradually at each stage. Delighted customers tell others, bringing new customers back to the top. Click a stage to see its metrics and prompt.
Prompt Guideline: Data to Prepare for AI to Analyse AARRR
If you only tell AI "do AARRR for my dessert shop", you get generic theory and possibly made-up rates. Prepare these 8 items first. The tags in the last column show which stage each item feeds. You do not need everything; AI can help you plan how to collect what is missing.
| Data to prepare | Examples of what to include | Where to find it | Used for stage |
|---|---|---|---|
| Goals and target customer | What you sell, to whom, and where you want to grow now (if you have done the 3Cs or Marketing Mix, paste the results) | Where to find itThe 3Cs Model and Marketing Mix articles in this knowledge hub | |
| Channels where people find you | Reach, link clicks and new LINE OA friends by channel, with the month collected | Where to find itMeta Business Suite, TikTok Analytics, LINE OA Manager | |
| Spend on getting customers | Ad spend, KOL fees and content production costs, in the same period as the numbers above | Where to find itMeta Ads Manager, freelancer receipts | |
| Definition of first experience | What counts as becoming a customer, such as messaging, asking for the menu or the first order, and how many people did it | Where to find itLINE OA chats, order list | |
| Purchase history per customer | How many times each customer bought and how many days apart (use codes instead of names), number of LINE OA blocks | Where to find itOrder book, POS system, LINE OA Manager | |
| Referral data | New customers who said they heard from a friend, number of reviews, shares or referral codes used | Where to find itQuestion at checkout, reviews on your page | |
| Sales, costs and fees | Monthly sales, average order value, cost per order, delivery and channel fees | Where to find itAccounting software, platform reports | |
| Why customers do not buy or do not come back | Common questions or complaints, people who message and then go silent, reasons customers give when they stop buying | Where to find itCustomer chats, a short survey |
6 rules before you paste data into AI
- 01Define each stage clearly firstWrite down what your business counts as Acquisition or Activation and include that definition in every prompt, so AI does not interpret it on its own
- 02Use numbers from the same periodEvery stage must come from the same month or week. Mixing periods makes the calculated rates wrong
- 03Do not let AI add benchmarksGood rates vary hugely by business. Tell AI to compare with your own previous month, not with averages that have no source
- 04Split by channelCustomers from TikTok and from friends often come back at different rates. Split the numbers by channel where you can
- 05Remove personal dataUse customer codes instead of names and phone numbers before pasting purchase history or chats into AI
- 06Fix one leak at a timePick the single biggest leak, test a fix and measure it before moving to the next one
No data yet? Let AI interview you first
AI helps define each stage for your business, then asks for the numbers one at a time and tells you how to start collecting what you do not have yet
Act as a growth consultant using Dave McClure's AARRR funnel (Acquisition, Activation, Retention, Referral, Revenue). Interview me ONE question at a time, in Thai. First help me define what each stage means for my business, then ask for the numbers of each stage for the same month. If I do not track something, tell me the cheapest way to start tracking it and move on. When done, show my funnel as a table with my definitions and numbers, and list what is still missing. Do not give recommendations yet.
Have data? Analyse the whole funnel
Calculate the rate between stages, build a funnel table and find the biggest leak compared with your own numbers
Act as a growth consultant. Analyse my AARRR funnel using ONLY the data below (all numbers from: [MONTH]). Business and target customer: [PASTE] 1. Acquisition: [HOW YOU COUNT IT] = [NUMBER], by channel: [PASTE], spend: [AMOUNT] 2. Activation: [HOW YOU COUNT IT] = [NUMBER] 3. Retention: [E.G. BOUGHT AGAIN WITHIN 30 DAYS] = [NUMBER] 4. Referral: [E.G. NEW CUSTOMERS FROM FRIENDS] = [NUMBER] 5. Revenue: sales [AMOUNT], orders [NUMBER], profit per order [AMOUNT] Last month's numbers for comparison (optional): [PASTE] Calculate the conversion rate between each stage, show the funnel as a table, find the biggest leak compared with my own previous numbers, explain likely causes (marked as guesses) and what data would confirm them. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
The 5 AARRR Stages with Metrics for Small Shops
Each stage has a key question, metrics a small shop can actually collect and a prompt to analyse it. The example is a homemade Thai dessert shop selling online through LINE OA, the same fictional shop as in the BMC, SWOT, 3Cs and Marketing Mix articles.
1Get customersAcquisition
Acquisition means getting the right people to know you and come to you. High views are not enough; they need to be people who could become customers. A small shop should know which channel brings the most right people at an acceptable cost.
- Which channel do paying customers come from most?
- Which channel brings many people but no buyers?
Review the ACQUISITION stage of my AARRR funnel. My business and target customer: [WHAT YOU SELL, TO WHOM, WHERE] How I count acquisition: [E.G. NEW LINE OA FRIENDS] New people per channel and spend per channel (same month): [PASTE] Calculate cost per new person for each channel from my numbers only, tell me which channel brings people who later buy, and suggest one channel test for next month. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
2First experienceActivation
Activation is when people who came in try the product and see its value for the first time. Online shops often lose customers here, for example slow replies, a hard-to-read menu or a difficult ordering process. Decide for yourself what a good first experience means for your business.
- Where do people who came in but did not order get stuck?
- What kind of first experience leads people to keep ordering?
Review the ACTIVATION stage of my AARRR funnel. My business and target customer: [WHAT YOU SELL, TO WHOM, WHERE] My definition of activation: [E.G. FIRST ORDER] New people and how many activated (same month): [PASTE] Questions and complaints from people who did not buy: [PASTE] Find where new people get stuck, suggest 3 fixes to the first experience, and how to measure each. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
3Come backRetention
Retention means getting customers to buy or use again. It shows whether the product is truly good. If customers do not come back, pouring money into new customers is like filling a leaky bucket.
- How many days until customers buy again?
- What do customers who do not return have in common?
Review the RETENTION stage of my AARRR funnel. My business and target customer: [WHAT YOU SELL, TO WHOM, WHERE] Order history per customer, using customer codes not names (with dates): [PASTE] LINE OA blocks per month: [PASTE] Calculate my repeat rate and typical time between orders from this data only, describe customers who do not return, and suggest a simple re-engagement plan that does not rely on constant discounts. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
4ReferReferral
Referral is when customers like you enough to tell others, recommend friends, review or share. It is a low-cost, trusted way to get new customers, but it only happens once customers are truly delighted.
- How many new customers say they heard from a friend?
- Why do customers recommend you, and can you make it easier?
Review the REFERRAL stage of my AARRR funnel. My business and target customer: [WHAT YOU SELL, TO WHOM, WHERE] How new customers found me (answers or counts, same month): [PASTE] Reviews and shares: [PASTE] Tell me why customers recommend me, what stops them, and design one simple referral idea I can run in LINE OA. Do not promise results; tell me what to track. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
5Earn revenueRevenue
Revenue is the income and profit from customers, not just total sales. Check whether the money from one customer is worth what you spent to get that customer.
- What is the real profit per order?
- Does one customer generate enough profit to cover the cost of getting them?
Review the REVENUE stage of my AARRR funnel. My business and target customer: [WHAT YOU SELL, TO WHOM, WHERE] Monthly sales, number of orders, cost per order, delivery and channel fees (with months): [PASTE] Spend to acquire customers in the same period: [PASTE] Calculate average order value, profit per order and a rough CAC versus customer value using ONLY my numbers. If data is missing, list what to collect instead of guessing. Then suggest 3 ways to raise profit per customer. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
Find the Leaks in Your Funnel
Divide the number of people at each stage by the stage before to get the pass-through rate. A stage with an unusually low rate compared with your own previous month is the one to fix first. The example below uses made-up numbers only to show the calculation; they are not benchmarks.
In this made-up example, 150 of 1,000 new friends placed a first order, or 15%, the lowest of all stages. So the shop should fix the first experience first, such as chat reply speed and ease of ordering. The Revenue stage is measured in money and profit, not as a rate.
Ask AI to find the cause of the leak
Focus on the single leakiest stage, find possible causes and design a small experiment
Here is my AARRR funnel with stage-to-stage rates: [PASTE TABLE] Customer feedback and chat notes about the leaking stage: [PASTE] Focus on the ONE stage with the biggest leak. List 5 possible causes, rank them by how easy they are to test, and design one small experiment for the top cause: what to change, for how long, and which number decides success. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
Make a 30-day growth plan
One main experiment for the leak and small tasks for other stages, with numbers to track
Here is my AARRR analysis: [PASTE] My team: [PEOPLE AND HOURS PER WEEK] Budget limit: [AMOUNT OR "SMALL"] Make a 30-day growth plan: one main experiment for the leaking stage, one small action for each other stage, a weekly checklist, and one number to track per action. Leave cost cells empty for me to fill. Answer in Thai, in short bullet points a beginner can act on. For every point, say which data it comes from, mark anything that is a guess, and never invent numbers, rates or industry benchmarks.
The Order of Stages Is Not Fixed
AARRR follows the journey of app or software users, who often try for free before paying. In many shops, customers pay on the very first purchase, so revenue comes before referral, and some sources put Revenue before Referral. Use the order that fits your customer journey; what matters is measuring all 5 areas.
Another idea is RARRA, which reorders the stages as Retention, Activation, Referral, Revenue and Acquisition, to stress keeping existing customers before spending more on new ones.
Once you know the leak, use OKRs to set a goal to fix it this quarter, adjust product, price or channels with Marketing Mix 4P / 7P and check that you are targeting the right customers with 3Cs Model
Mistakes beginners often make
- Watching only views or likes, which do not show whether the business is really growing
- Pouring money into new customers while existing customers do not come back
- Not defining each stage clearly, so the counts differ every month
- Comparing numbers with benchmarks from the internet or from AI instead of your own numbers
- Fixing every stage at once, so you cannot tell what actually worked
Checklist before you use it
- Each stage is defined for your business and written down so the team counts the same way
- Numbers for every stage come from the same period
- Compared with your own previous month, not benchmarks without a source
- One biggest leak chosen, with an experiment and a number that decides success
- Customer codes used instead of personal data before pasting into AI
- A monthly funnel review is scheduled
References: Wikipedia: Dave McClure (Startup Metrics for Pirates, 2007), Shopify: AARRR Metrics and RARRADave McClure, Startup Metrics for Pirates, Ignite Seattle (2007) · The example business and numbers in this article are fictional and are not benchmarks. AI analysis is a suggestion to consider; check it against real data before making investment decisions.
Share it with a friend who is starting a business. Share to Facebook Share on LINE


