AI Can Tell the Future: Forecasting with AI

In today's newsletter:

• Latest Podcasts

  • AI Can Tell the Future: Forecasting with AI

  • The AI Forecasting Framework

  • Today's 10-Minute Action Plan

• Premium: The AI Forecasting Toolkit: Predict Reven …

AI Can Tell the Future: Forecasting with AI

Your CFO asks: "What's our revenue going to be next quarter?"

You guess. "Uh... $150K? Maybe?"

They ask: "Based on what?"

You have no answer. You're just... hoping.

This is the forecasting blindness problem. You're making decisions with no idea what's coming.

The fix? AI-powered forecasting: using historical data and machine learning to predict revenue, churn, demand, and cash flow.

The AI Model That Predicted a $50K Shortfall

Let me tell you about “Olivia”, founder of a 7-person SaaS company.

Olivia's forecasting method: Gut feel + hope.

Example conversation with her CFO:

CFO: "What's Q3 revenue going to be?"

Olivia: "We're at $40K MRR now. Growing 10% per month. So... $55K MRR by Q3? That's $165K for the quarter."

CFO: "Are you accounting for churn?"

Olivia: "Uh... sure?"

What actually happened in Q3:

• Revenue: $120K (not $165K)

• $45K shortfall

• Reason: Higher-than-expected churn + slower growth

Olivia almost ran out of cash.

Then Olivia implemented an AI forecasting model.

What she did:

1. Gathered 12 months of historical data:

• Monthly revenue

• New customers

• Churned customers

• Customer lifetime value (LTV)

• Marketing spend

2. Fed it into an AI forecasting tool (ChatGPT + Google Sheets script)

3. Asked the AI:
"Based on the last 12 months, predict revenue for the next 3 months. Account for churn and seasonal trends."

AI output:

• Month 1: $42K (±$3K)

• Month 2: $44K (±$3K)

• Month 3: $46K (±$4K)

• Q3 total: $132K

Actual results:

• Month 1: $43K

• Month 2: $45K

• Month 3: $44K

• Q3 total: $132K

AI forecast accuracy: 99%

Key insight: "I used to think forecasting was guesswork. But AI turned my historical data into a crystal ball. Now I make decisions based on predictions, not hope."

Why Founders Don't Forecast (And Why That's Dangerous)

Here's why most microteam founders avoid forecasting:

1. "I don't have enough data"

• You think you need years of data

• Reality: 6-12 months is enough for basic forecasting

2. "Forecasting is for big companies"

• Wrong. Small companies need forecasting more (less margin for error)

3. "I don't know how to build models"

• You don't need to. AI does it for you.

4. "My business is too unpredictable"

• Even volatile businesses have patterns

• AI finds them

Think of forecasting like weather prediction.

Without forecasting:

• You don't know if it's going to rain

• You leave your umbrella at home

• You get soaked

With forecasting:

• You see rain is 80% likely

• You bring an umbrella

• You stay dry

AI forecasting is your business weather report.

Why This Matters for Microteams

Big companies have data science teams building forecast models.

You? You're flying blind.

Here's why AI forecasting is critical:

• Prevent cash crunches. Know when revenue will dip before it happens.

• Hire confidently. Only add headcount if the forecast supports it.

• Plan inventory. Predict demand spikes (e-commerce, physical products).

• Set realistic goals. Stop guessing, start predicting.

• Spot problems early. If AI predicts a revenue drop, you can act now.

Stop hoping, start forecasting.

The AI Forecasting Framework

Here's how to use AI to predict revenue, churn, demand, and cash flow.

Step 1: Identify What to Forecast

What metrics matter most to your business?

Common forecasting targets:

Pick 1-2 metrics to start.

Olivia's choice: Monthly revenue + churn rate.

Step 2: Gather Historical Data

AI needs data to make predictions.

Minimum data needed:

• 6 months of historical data (bare minimum)

• 12 months is better

• 24+ months is ideal (captures seasonality)

What data to collect:

For revenue forecasting:

• Monthly revenue (past 12 months)

• New customers per month

• Churned customers per month

• Average revenue per customer

For churn forecasting:

• Monthly churn rate (past 12 months)

• Customer cohort data (when did they sign up?)

For demand forecasting:

• Units sold per month (past 12-24 months)

• Marketing spend per month

• Seasonal events (holidays, promotions)

Export this data to a CSV or Google Sheet.

Step 3: Use AI to Generate Forecasts

You don't need to build a model from scratch. Use AI tools.

Option 1: ChatGPT + Google Sheets (easiest, free)

Step-by-step:

1. Prepare your data in Google Sheets

2. Copy the data and paste into ChatGPT

Prompt:

"Here is 12 months of revenue data for my SaaS business. Predict revenue for the next 3 months. Account for churn trends and growth rates.

[Paste data]

Output: Predicted revenue for each of the next 3 months."

3. ChatGPT generates the forecast

Output:

Based on your historical data:

• Month 13: $42K (±$3K)

• Month 14: $44K (±$3K)

• Month 15: $46K (±$4K)

Assumptions:

• Average growth rate: 5% per month

• Churn rate: 8% per month

• Seasonal trends: None detected

4. Use this to plan

Option 2: Google Sheets FORECAST function (built-in)

Formula:

=FORECAST(future_month, known_revenues, known_months)

Example:

• Known data: Months 1-12, Revenue $30K-$50K

• Forecast for Month 13: =FORECAST(13, B2:B13, A2:A13)

This gives a simple linear projection.

Pros: Free, instant, no AI needed

Cons: Less sophisticated (doesn't account for churn, seasonality)

Option 3: Specialized AI Forecasting Tools

Tools:

For most microteams: Start with ChatGPT + Google Sheets (free, easy).

Step 4: Validate the Forecast

Don't blindly trust AI. Test it.

Back-test:

1. Use historical data from Months 1-92. Ask AI to predict Months 10-123. Compare AI's prediction to actual results

If AI is within 10-20% accuracy → Good enough.

If AI is way off → Check your data or add more context (seasonality, external events).

Olivia's back-test:

• AI predicted Month 10: $38K

• Actual Month 10: $39K

• Accuracy: 97%

This gave her confidence to use AI forecasts for planning.

Step 5: Update Monthly

Forecasts aren't static. Update them as new data comes in.

Every month:

1. Add the latest month's data to your spreadsheet

2. Re-run the AI forecast

3. Compare the prediction to actuals (learn from errors)

4. Adjust your plans based on the updated forecast. This keeps your forecast accurate and actionable.

Step 6: Use Forecasts to Make Decisions

Forecasting is pointless if you don't act on it.

Example decisions:

Revenue forecast shows a dip next quarter:

• Action: Cut non-essential expenses now

• Action: Launch a promotion to boost revenue

Churn forecast shows increasing churn:

• Action: Implement retention campaign

• Action: Interview churned customers to find root cause

Demand forecast shows spike in December:

• Action: Stock up on inventory in November

• Action: Hire temp workers for the rush

Cash runway forecast shows 4 months left:

• Action: Cut burn rate or raise capital now (not in 3 months when it's too late)

Forecasts give you time to act proactively, not reactively.

AI Forecasting Examples by Business Type

SaaS:

• Forecast: Monthly MRR, churn rate

• Use case: Plan hiring, identify at-risk revenue

E-commerce:

• Forecast: Units sold per product, demand spikes

• Use case: Inventory planning, production schedules

Consulting / Agency:

• Forecast: Monthly revenue, client acquisition

• Use case: Cash flow planning, capacity planning

Content / Media:

• Forecast: Pageviews, ad revenue

• Use case: Content strategy, ad budget planning

Common Forecasting Mistakes

Mistake 1: Not enough data

• You need at least 6 months

• Ideally 12-24 months

Mistake 2: Ignoring external events

• Seasonality (holidays, summer slump)

• Market changes (competitor launches, economic shifts)

• Add context to your AI prompt

Mistake 3: Treating forecasts as guarantees

• Forecasts are probabilities, not certainties

• Always include a margin of error (±10-20%)

Mistake 4: Not updating regularly

• Old forecasts are useless

• Update monthly with fresh data

Mistake 5: Forecasting but not acting

• If the forecast shows a problem, fix it now

• Don't just watch the crash coming

Advanced: Scenario Planning with AI

Don't just forecast one future. Forecast three.

Scenario planning:

Best case: Everything goes right (20% growth, low churn)

Base case: Most likely outcome (10% growth, normal churn)

Worst case: Everything goes wrong (0% growth, high churn)

ChatGPT prompt:

"Based on this data, create 3 forecasts for the next 3 months:

1. Best case (aggressive growth, low churn)

2. Base case (moderate growth, normal churn)

3. Worst case (flat growth, high churn)"

Use this to plan for multiple futures.

Today's 10-Minute Action Plan

You don't need to build a complex forecast model today. Just make one simple prediction.

Here's what to do in the next 10 minutes:

1. Open Google Sheets

2. Enter the last 6 months of revenue (one column: Month, one column: Revenue)

3. Paste the data into ChatGPT and ask:
   "Predict revenue for the next 3 months based on this data."

4. Write down the prediction

5. Set a reminder for next month to compare prediction vs. actual

That's it. One forecast created, 10 minutes.

Next month, refine it. In 3 months, you'll know how accurate AI forecasting is for your business.