Most small business owners are flying partially blind. They have a gut feeling about what's working but haven't confirmed it with data. They know roughly how much revenue they're making but not which products, clients, or channels drive the most of it. They spend on marketing but aren't sure which spend is returning anything.
The data to answer all of these questions already exists — in your accounting software, your payment processor, your spreadsheets, your email tool, your social analytics. The problem isn't a lack of data. It's a lack of a system for turning it into decisions.
This playbook builds that system using AI. No SQL, no Python, no Tableau. Just structured prompts that take your raw numbers and return clear insights about what's working, what isn't, and what to do next.
✅ What you'll have when done: A data source map, a monthly business scorecard with 8–10 KPIs, an AI revenue trend analyzer, a customer segmentation framework, a marketing performance analyzer, and a monthly AI review ritual — all prompt-based and ready to run every month in under 30 minutes.
From Raw Data to Clear Decision — The AI Analysis Flow
Raw Data
Numbers from your tools
Structured Prompt
You paste + provide context
AI Analysis
Pattern recognition + interpretation
Insight
What the numbers mean
Decision
What to do next
The Data You Already Have
Before building any analysis, you need to know what you're working with. Most small businesses have data in 6 places — each one answers different questions:
Accounting / Bookkeeping
Revenue, expenses, profit margin, cash flow
Payment Processor
Transaction history, average order value, refund rate
CRM / Customer List
Purchase frequency, lifetime value, churn
Email Marketing
Open rates, click rates, subscriber growth, revenue per email
Social / Ads
Reach, engagement, ad spend, cost per click, conversions
Website Analytics
Traffic, top pages, conversion rate, traffic sources
The 10 KPIs Every Small Business Should Track
You don't need 50 metrics. You need the right 10. These are the indicators that give you an accurate picture of business health every single month:
Monthly Revenue
Total income this month vs. last month vs. same month last year
Revenue per Customer
Total revenue ÷ number of paying customers this month
New vs. Returning
% of revenue from new customers vs. repeat buyers
Churn Rate
% of customers who didn't buy again (for recurring businesses)
Gross Margin
(Revenue − Cost of Goods) ÷ Revenue × 100
Net Profit
Revenue minus all expenses — the actual number that matters
Customer Acquisition Cost
Total marketing spend ÷ new customers acquired
Marketing ROI
Revenue attributed to marketing ÷ marketing spend
Average Response Time
How long customers wait for a reply or fulfilment
Month-over-Month Growth
(This month − Last month) ÷ Last month × 100
⚠️ Affiliate disclosure: Some links on this page may be affiliate links. We may earn a commission at no cost to you. We only recommend tools we actively use and test.
The 6-Step Playbook
Audit Your Data Sources
20 min · Claude · One-time setup — know what you have before you analyze it
Before running any analysis, you need a clear picture of where your data lives, how reliable it is, and what questions it can actually answer. Most small businesses have pockets of good data they've never connected to business decisions — and gaps they don't know exist.
This prompt creates a data map that becomes the foundation for every analysis in this playbook:
You are a business analyst helping me understand what data I have and what decisions it can support. My business: [name and what I sell] Business model: [e.g. one-time product sales, monthly subscription, project-based services, recurring retainer] In business for: [e.g. 2 years] My current data sources (describe what you actually have access to): 1. Accounting: [e.g. QuickBooks with 18 months of transactions / just a spreadsheet / nothing structured] 2. Customer data: [e.g. spreadsheet of 200 customers with purchase history / CRM / just emails] 3. Payment data: [e.g. Stripe dashboard going back to 2023 / PayPal statements / manual invoices in a folder] 4. Website: [e.g. Google Analytics set up / no tracking / Plausible with 6 months of data] 5. Email marketing: [e.g. Mailchimp with 800 subscribers, 18 months of campaign data / none] 6. Social / ads: [e.g. Meta Ads Manager, $500/mo spend, 3 months of data / no paid ads] 7. Other: [anything else — e.g. a Google Sheet tracking jobs, a booking system, a POS] My biggest business questions right now (what do you most want to know?): 1. [e.g. Which products are actually most profitable?] 2. [e.g. Is my marketing spend working?] 3. [e.g. Why did revenue drop in Q4?] Based on this, create my Data Source Map: 1. DATA INVENTORY — For each source I listed: - Quality rating (1–5 stars) based on completeness and reliability - What questions it can answer - What questions it cannot answer from this source alone 2. DATA GAPS — What important questions I have no data to answer yet, and the simplest way to start collecting that data 3. QUICK WINS — 3 analyses I could run right now from existing data that would answer high-value business questions 4. DATA COLLECTION PRIORITIES — The 2–3 data sources I should set up or improve first, ranked by business impact 5. ANALYSIS CALENDAR — A suggested monthly analysis schedule: what to review weekly, monthly, and quarterly, based on my business model
Build Your Monthly Business Scorecard
20 min to set up · 10 min/month to run · Claude · Your business health pulse
The scorecard is the centrepiece of your data practice. Once a month, you fill in 8–10 numbers and run a single AI prompt that interprets them — flagging what's healthy, what's concerning, and what deserves attention this month.
The first run of this prompt also designs your scorecard itself, tailored to your business model:
You are a business analyst helping me build and interpret a monthly business scorecard. My business: [name, what I sell, business model] Industry / type: [e.g. service business, e-commerce, SaaS, consulting, retail] Stage: [e.g. early (under $100k/yr), growing ($100k–$500k/yr), established (over $500k/yr)] PART 1 — DESIGN MY SCORECARD: Based on my business type and stage, design a monthly scorecard with exactly 8–10 KPIs. For each KPI: - Name and definition (how to calculate it) - Where to get this number (which tool or source) - Why it matters for my specific business model - What a healthy target range looks like for a business at my stage - What a warning sign looks like Format as a table I can use as a monthly template. PART 2 — INTERPRET THIS MONTH'S NUMBERS: (Fill this in each month after designing the scorecard) Month: [Month Year] My scorecard numbers this month: [List each KPI and its value — e.g. "Monthly Revenue: $12,400"] Previous month for comparison: [e.g. "Monthly Revenue: $10,800"] Same month last year (if available): [e.g. "Monthly Revenue: $8,200"] Any context I should know: [e.g. ran a sale in week 2, lost a major client, launched a new product] Analyze these numbers and give me: 1. HEADLINE — One sentence: is this month good, concerning, or mixed? 2. WHAT'S WORKING — Top 2–3 positive signals with brief explanation 3. WHAT NEEDS ATTENTION — Top 2–3 warning signs or areas to watch 4. ROOT CAUSE ANALYSIS — For any metric that's off, what's the most likely explanation? 5. THIS MONTH'S #1 PRIORITY — The single most important thing to focus on based on this data 6. 3 ACTIONS — Specific, time-bound actions to take this month based on what the data shows
💡 Run Part 1 once, Part 2 every month. After your first run, save Part 2 as a recurring template. Each month: fill in the numbers, paste in the previous month's for comparison, add any relevant context, run it. Takes 10 minutes. Gives you a board-quality review of your own business.
Analyze Your Revenue Patterns
30 min · Claude · Find the trends hiding in your numbers
Revenue data in isolation just tells you how much you made. Revenue data with pattern analysis tells you why — which months are consistently strong, whether you're growing or plateauing, which products or services are driving the most revenue, and where the growth is actually coming from.
Export your revenue history from your accounting tool or payment processor and paste it directly into this prompt:
You are a financial analyst helping me understand the patterns in my business revenue data. My business: [name and what I sell] Business model: [e.g. project-based, subscription, one-time product sales, mix] My revenue data (paste directly from your spreadsheet or accounting export): [Format: Month | Revenue | Number of transactions | Average transaction value] [Example: Jan 2025 | $8,200 | 14 | $586 Feb 2025 | $7,400 | 12 | $617 Mar 2025 | $11,800 | 19 | $621 ... continue for all available months] If you have revenue broken down by product/service, include that too: [Product A: Jan $3,200 | Feb $2,800 | Mar $4,100 ...] [Product B: ...] Analyze this data and provide: 1. OVERALL TREND — Is revenue growing, flat, or declining? Calculate month-over-month average growth rate. 2. SEASONALITY ANALYSIS — Are there consistent patterns by month or quarter? Which months are reliably strong or weak? 3. GROWTH CONTRIBUTORS — If revenue is growing, where is the growth coming from? More transactions? Higher average value? Both? 4. ANOMALIES — Flag any months that were significantly above or below trend. What might explain them? 5. PRODUCT / SERVICE MIX (if data provided) — Which offering drives the most revenue? Which has the highest growth rate? Which is declining? 6. REVENUE VELOCITY — At the current trajectory, what might the next 3 months look like? (Range, not a precise prediction.) 7. THE ONE THING — Based purely on this revenue data, what is the single most important thing this business should focus on to accelerate growth? Present findings clearly — assume I am not a financial analyst. Use plain language.
ℹ️ Don't have clean monthly data? Pull your payment history from Stripe, PayPal, or your bank statement. Export as CSV and upload it directly to Claude or GPT-5.4 — both can read and aggregate raw transaction data. Even 6 months of data is enough to spot meaningful patterns.
Identify Your Best and Worst Customers
25 min · Claude · Focus your energy where it returns the most
Not all customers are created equal. The 80/20 rule applies in most small businesses — roughly 20% of customers typically generate 80% of revenue. But more importantly, some customers are significantly more profitable, more loyal, and more likely to refer others. Knowing who they are changes how you allocate your sales and marketing effort.
The 4 Customer Segments
| Segment | Definition | AI Strategy |
|---|---|---|
| ⭐ VIP | High spend, high frequency, long tenure. Your top 10–20%. | Protect, nurture, ask for referrals. Never lose one by neglect. |
| 📈 Growth | Recent buyers with increasing purchase frequency. Rising stars. | Upsell, cross-sell, deepen the relationship. These are tomorrow's VIPs. |
| ⚠️ At Risk | Previously regular customers who've gone quiet. Haven't bought in X months. | Re-engagement campaign. AI writes a personalised win-back message. |
| 💤 Low Value | Bought once, low spend, low engagement. High effort, low return. | Automate their journey. Don't invest manual time here. |
You are a customer analytics specialist helping me segment my customer base by value. My business: [name and what I sell] Customer relationship type: [one-time purchases / recurring / project-based / subscription] My customer data (paste from your CRM, spreadsheet, or accounting export): [Format: Customer Name/ID | Total Revenue | Number of Purchases | First Purchase Date | Last Purchase Date | Average Order Value] [Example: Bloom Bakery | $8,400 | 6 | Jan 2024 | Dec 2025 | $1,400 Torres Plumbing | $3,200 | 2 | Mar 2025 | Jun 2025 | $1,600 Chen Consulting | $12,800 | 9 | Oct 2023 | Feb 2026 | $1,422 ...] Analyse this customer data and provide: 1. SEGMENTATION — Assign each customer to a segment: - VIP (top 20% by lifetime value + recency) - Growth (recent + increasing frequency) - At Risk (previously active, now quiet — define "quiet" based on my data) - Low Value (low spend, infrequent) 2. SEGMENT SUMMARY TABLE — How many customers in each segment, total revenue per segment, % of total revenue 3. VIP PROFILE — Describe my best customers: what do they have in common? (industry, size, purchase pattern, AOV) 4. AT-RISK ALERT — List any at-risk customers by name with their last purchase date and estimated revenue at risk 5. REVENUE CONCENTRATION RISK — What % of my revenue comes from my top 3 customers? Is this a risk? 6. ACQUISITION INSIGHT — Based on my VIP profile, what type of new customer should I be prioritising in sales and marketing? 7. RECOMMENDED ACTIONS — One specific AI-powered action for each segment (what to send, when, and why)
⚠️ Revenue concentration is a real risk. If your top 3 customers represent more than 50% of your revenue, you have a structural vulnerability. The analysis will flag this. The fix is systematic: use your VIP profile to attract more customers who look like your best ones, while deepening relationships with current VIPs to reduce churn risk.
Build a Marketing Performance Tracker
25 min · GPT · Know what's working before you spend another dollar
Most small businesses run marketing on intuition — boosting posts that "feel" like they're working, continuing email campaigns without knowing if they're driving revenue, running ads without tracking whether they're profitable. This prompt replaces intuition with a structured performance analysis.
Run it monthly with your actual marketing data to know exactly where to cut and where to invest more:
You are a marketing analyst helping me understand which parts of my marketing are working. My business: [name and what I sell] My average customer lifetime value (LTV): [e.g. $1,200 — total revenue per customer over their time with you] My average profit margin: [e.g. 60%] My marketing channels and this month's data: CHANNEL 1 — [e.g. Email Marketing]: - Platform: [Mailchimp / Kit / other] - Subscribers: [total list size] - Emails sent: [number this month] - Open rate: [%] - Click rate: [%] - Revenue attributed to email (if tracked): [$] - Monthly cost: [$] CHANNEL 2 — [e.g. Instagram / Organic Social]: - Platform: [Instagram / Facebook / LinkedIn / TikTok] - Followers: [number] - Posts this month: [number] - Average reach per post: [number] - Engagement rate: [%] - Leads or sales attributed: [number or $, if tracked] - Monthly time cost: [hours] CHANNEL 3 — [e.g. Paid Ads]: - Platform: [Meta / Google / other] - Monthly spend: [$] - Impressions: [number] - Clicks: [number] - Cost per click: [$] - Conversions (leads or sales): [number] - Cost per conversion: [$] - Revenue attributed: [$] CHANNEL 4 — [e.g. Referrals / Word of Mouth]: - New customers from referrals this month: [number] - Revenue from referrals: [$] - Any referral program cost: [$] [Add more channels as needed. Skip channels you don't use.] Analyse this marketing data and provide: 1. CHANNEL PERFORMANCE SCORECARD — Rate each channel: High ROI / Moderate / Low / Unknown (needs tracking) 2. CUSTOMER ACQUISITION COST BY CHANNEL — Calculate CAC for each channel where data allows. How does each compare to my LTV? 3. HIGHEST VALUE CHANNEL — Which channel is generating the best return? Why? 4. WORST PERFORMING CHANNEL — Which should I cut or pause? What would I save/reallocate? 5. ATTRIBUTION GAPS — Where am I spending without tracking return? How to fix this with minimal effort? 6. 3 RECOMMENDATIONS — Specific, actionable changes to my marketing mix this month based on this data, ranked by expected impact 7. $100 EXPERIMENT — If I had an extra $100 to test next month, where should I put it based on what this data suggests?
💡 The single highest-ROI marketing channel for most small businesses is referrals. Zero cost, highest trust, highest conversion rate. If your data shows referral customers exist but you have no formal referral program, that's your #1 gap. A well-written referral ask (AI can draft one in 60 seconds) often outperforms months of paid advertising.
Create Your Monthly AI Business Review
30 min/month · Claude · Turns all your data into one clear decision
Once your scorecard, revenue analysis, customer segments, and marketing data are assembled, they need to be read together — not as separate reports, but as a unified picture of your business. The monthly review prompt does exactly that.
Run it on the first Monday of every month with the previous month's data. It replaces the role of a business advisor for routine monthly review:
You are my business advisor helping me run a structured monthly review. Business: [name and what I sell] Review month: [Month Year] SCORECARD SUMMARY (from Step 2): [Paste your 8–10 KPIs and their values — current month vs. previous month] REVENUE HIGHLIGHTS (from Step 3): - This month's revenue: [$] - vs. last month: [+/- % change] - vs. same month last year: [+/- % change] - Notable: [any anomaly or context] CUSTOMER SNAPSHOT (from Step 4): - New customers acquired: [number] - Repeat purchases: [number] - Any customers moved to At Risk this month: [names or "none"] - Any VIP customer activity worth noting: [e.g. "Chen Consulting placed their largest order yet"] MARKETING SUMMARY (from Step 5): - Best performing channel this month: [channel + why] - Anything cut or paused: [channel + reason] - Total marketing spend: [$] - Estimated marketing-attributed revenue: [$] CONTEXT: - What was my main focus this month? [e.g. launched new service, hired someone, ran a sale] - What went unexpectedly well? - What went unexpectedly badly? - Biggest distraction or time sink this month? Now run my monthly review: 1. MONTH IN ONE SENTENCE — How did this month go overall? 2. TOP 3 WINS — What actually worked? What should I repeat or double down on? 3. TOP 3 CONCERNS — What needs attention? What pattern is emerging that could become a problem if ignored? 4. THE NUMBER THAT MATTERS MOST — Out of everything I've shared, which single metric most accurately reflects where this business is heading? 5. NEXT MONTH'S PRIORITIES — 3 specific, focused priorities for next month, based solely on this data. (Not a to-do list — strategic priorities.) 6. ONE QUESTION I SHOULD BE ASKING — Based on this data, what question am I NOT asking that I should be? What would the answer change?
ℹ️ Block 90 minutes on the first Monday of every month. 30 minutes to gather and fill in the numbers, 30 minutes for the AI review, 30 minutes to turn the priorities into actual calendar items and tasks. This is one of the highest-leverage recurring activities a small business owner can do. Most don't do it at all — which is exactly why the ones who do pull ahead.
What Breaks Most Small Business Data Practices
- Waiting until tax time to look at the numbers. Looking at data quarterly or annually is too slow to act on. By the time you notice a trend in an annual review, you've had months of suboptimal decisions. Monthly is the minimum; weekly pulse checks on 2–3 key metrics is ideal.
- Tracking vanity metrics instead of business metrics. Instagram followers and website visitors feel good but don't pay bills. Track the numbers that connect to revenue: conversion rate, customer acquisition cost, average order value, repeat purchase rate. If it doesn't link to money, it's a secondary metric.
- Analysis without action. The point of data analysis is a decision. Every review session should end with at least one specific change — to a product, a marketing channel, a pricing strategy, a customer segment focus. If you run the analysis and nothing changes, the analysis was entertainment, not work.
- Not tracking marketing attribution. "I don't know where my customers come from" is one of the most expensive things a small business owner can say. Even a simple "How did you hear about us?" question at checkout tells you more than months of social media analytics.
- Ignoring the customer segment data. Most small businesses treat all customers equally — same email, same offer, same attention. Segmentation shows you that your VIP customers deserve different treatment, your at-risk customers need a specific intervention, and your low-value customers should be served efficiently rather than lavishly.
Playbook Summary — Your Data Dashboard Checklist
- Data source audit completed — know what you have and what's missing
- Data gaps identified and 2–3 collection improvements prioritized
- Business scorecard designed — 8–10 KPIs tailored to your business model
- Scorecard template saved and first month's numbers entered
- Revenue history exported and pattern analysis run
- Seasonality and growth trends identified
- Customer data exported and segmentation analysis run
- VIP profile defined — you know who your best customers look like
- At-risk customers identified by name with re-engagement actions assigned
- Marketing channel performance scorecard completed
- At least one underperforming channel identified for cut or pause
- Monthly business review scheduled on first Monday of each month
- First full monthly review completed — 3 strategic priorities set
Ready for the Final Playbook?
You've automated your sales, marketing, support, and data review. Playbook 6 ties it all together — building simple internal tools and workflows that run the operational side of your business without you having to be there.
Playbook 6: Internal Tools Builder → All PlaybooksFrequently Asked Questions
Do I need to know data analysis to use this playbook?
No. Every prompt is structured so that you paste in your raw numbers — even a simple copy-paste from a spreadsheet — and AI does the pattern recognition, interpretation, and recommendation writing for you. No statistics knowledge required.
What data do I need to start?
You probably have more than you think — monthly revenue from accounting software or bank statements, a customer list with purchase history, and some form of marketing data. Step 1 maps exactly what you have and identifies which 3 analyses to run first for the highest immediate value.
Can AI analyze data from a spreadsheet?
Yes. Copy rows directly from Google Sheets or Excel and paste them into Claude or ChatGPT. For larger datasets, export as CSV and upload directly — both tools can read, calculate, and interpret spreadsheet data without any special setup.
What if my numbers aren't that impressive?
That's exactly when AI data analysis is most valuable. Flat or declining numbers contain the most actionable information — they tell you where the problem is. AI is particularly good at identifying which specific segment, channel, or product is causing underperformance, so you can fix the right thing instead of guessing.
How often should I run AI data analysis?
Monthly for the full review (Step 6), weekly for a quick pulse on 2–3 key metrics. The monthly review takes about 30 minutes once you have the prompts set up. Businesses that review monthly consistently outperform those that only look at data at tax time.
