Published March 9, 2026

How to Build AI Sales Forecasting Using CRM and Billing Data

March 24, 2026

AI sales forecasting is a feature inside a SaaS product that predicts future revenue using CRM pipeline data and billing signals. It combines deal activity from systems like Salesforce or HubSpot with subscription and payment data from platforms like Stripe to produce more accurate forecasts.

To build this, your product needs access to both pipeline data and revenue data. CRM systems track deals and their progression, while billing platforms track subscriptions and payments. Each system has different APIs and schemas.

Unified provides category-specific APIs for CRM and Payments, allowing your product to retrieve pipeline and billing data through consistent objects. This makes it possible to build forecasting features without maintaining separate integrations for each provider.

Why SaaS Products Build AI Sales Forecasting Features

Many B2B SaaS products include forecasting capabilities for their customers.

Examples include:

Revenue intelligence platforms
Sales analytics tools
RevOps dashboards
Customer success platforms
AI sales assistants

These products help teams answer questions like:

'What revenue will close this quarter?'
'Which deals are most likely to close?'
'How does pipeline compare to actual revenue?'

Without combining CRM and billing data, forecasts are incomplete or inaccurate.

Common AI Sales Forecasting Use Cases

AI forecasting features typically support several workflows.

Revenue prediction

Estimate future revenue based on pipeline and historical performance.

Deal forecasting

Predict the likelihood of deals closing.

Pipeline analytics

Analyze deal stages, velocity, and conversion rates.

Forecast vs actual comparison

Compare expected revenue with billing data.

Risk detection

Identify deals or accounts that may not close.

Unified Categories Used

AI sales forecasting combines data from two categories.

Category	Description	Key Objects
CRM	Pipeline and account data from platforms like Salesforce and HubSpot	contact, company, deal, event
Payments	Subscription and transaction data from platforms like Stripe	payment, subscription

These categories provide the inputs needed for forecasting.

Unified Objects and Key Fields

CRM Objects

CRM data provides pipeline and deal context.

Object	Key Fields	Purpose
Deal	amount, probability, stage_id, closing_at	Forecast pipeline revenue
Contact	name, emails	Customer identity
Company	name, domains	Account grouping
Event	type, created_at	Activity history

Important fields for forecasting:

amount – deal value
probability – likelihood of closing
closing_at – expected close date
stage_id – pipeline stage

These fields allow your product to estimate expected revenue.

Payments Objects

Billing data provides actual revenue signals.

Object	Key Fields	Purpose
Payment	total_amount, currency, contact_id	Cash collected
Subscription	status, current_period_end_at, interval	Recurring revenue

Important subscription fields include:

status – active, canceled, paused
current_period_start_at / current_period_end_at – billing cycle
interval / interval_unit – billing frequency
lineitems – pricing details

These fields allow your product to calculate recurring revenue.

Mapping Data to Forecast Metrics

Forecasting models combine CRM and billing signals.

Pipeline value

weighted_pipeline = deal.amount * deal.probability

Estimates expected revenue based on deal probability.

Projected revenue

projected_revenue = sum(weighted_pipeline by closing_at period)

Forecasts revenue by time period.

Actual revenue

actual_revenue = sum(payment.total_amount)

Tracks collected revenue.

Recurring revenue

MRR = sum(active subscription monthly value)
ARR = MRR * 12

Measures predictable revenue streams.

Combining these metrics allows your product to compare forecasted vs actual performance.

Connecting Customer Data Sources

Customers connect their CRM and billing platforms using Unified Connect.

Typical process:

Your application launches the authorization flow.
The user selects integrations such as Salesforce, HubSpot, or Stripe.
The user authorizes access.
Unified returns a connection_id.

Your application stores:

user_id → connection_id

All API requests reference this identifier.

Retrieving CRM and Billing Data

Your application retrieves data using Unified endpoints.

CRM:

GET /crm/{connection_id}/deal
GET /crm/{connection_id}/contact
GET /crm/{connection_id}/company
GET /crm/{connection_id}/event

Payments:

GET /payment/{connection_id}/payment
GET /payment/{connection_id}/subscription

These endpoints return normalized objects across providers.

Common filters include:

updated_gte for incremental updates
contact_id for linking revenue to accounts

Building an AI Forecasting Pipeline

AI forecasting features typically follow this pattern.

1. Retrieve data

Fetch deals, payments, and subscriptions.

2. Aggregate metrics

Calculate pipeline, revenue, and activity metrics.

3. Train or apply models

Use historical data to predict outcomes.

4. Generate forecasts

Estimate revenue for future periods.

5. Store results

Persist forecasts in your application.

6. Display insights

Show forecasts in dashboards or reports.

This pipeline allows your product to deliver predictive insights.

Generating AI Forecast Insights

AI models can generate explanations alongside forecasts.

Example prompt:

{
  "messages": [
    { "role": "system", "content": "You are a sales forecasting assistant." },
    { "role": "user", "content": "Explain why revenue is projected to decline next quarter." }
  ]
}

The model can produce: