Improve forecasting with AI projections (BETA)

In HubSpot, you can use predictive artificial intelligence to project future sales based on closed won deals in the current period. The AI forecast provides an additional perspective you can use to calibrate an accurate forecast for your team. This information helps provide an overview for the whole organization, so all deals factor into these tables. 

Please note: this feature is in BETA, learn how to opt your account into the beta.

Set up AI forecasting

To set up the forecast AI forecasting or opt-out, you must be a Super Admin. To set up AI forecasting:

• In your HubSpot account, navigate to Reporting & Data > Forecast.
• Click the Analyze tab.
• Click Turn on AI forecasting.
• Then, click Yes, turn on AI forecasting to confirm you would like to turn on the feature.

After you opt-in to the feature, there is a waiting period of seven days or more while data is collected to make the most accurate predictions.

If you would like to opt-out of AI forecasting: 

• In your HubSpot account, click the settings settings icon in the top navigation bar.
• In your left sidebar menu, navigate to Objects > Forecast.
• In the AI Forecasting section click Turn off AI Forecasting.

Use AI forecasting

• In your HubSpot account, navigate to Reporting & Data > Forecast.
• Click the Analyze tab.
• Use the dropdown menus at the top to filter by pipeline.
• In the Overview for all teams table at the top, you can view the following:

• • Weighted pipeline value: the total Amount multiplied by the Deal probability of all deals including Closed won deals.

Please note: the deal stage property and the deal amount configured for your account are used to calculate the closed won value.

• • Closed Won: the total amount of closed won deals.
  • Forecast submission: the sum of forecast submissions for all teams. This metric will only be visible for accounts on a monthly cadence.

The AI forecast table at the top uses Closed won deal data. Specifically, the date a deal is moved to the deal stage Closed won and the amount of the deal. The AI forecast is provided as a range, with the midpoint of the range being where you are most likely to land. In the table, you can view the following:

• • Most likely: the midpoint of the range and where the model projects you are most likely to land.
  • Upper: the upper bound of the range you’re projected to land within.
  • Lower: the lower bound of the range you’re projected to land within.

Each projection is created using Closed won deal data. The accuracy of the projection depends on the accuracy of your data. You can help improve the accuracy of the projection by ensuring deal data is correct and up to date.

Please note: projections are estimates and are not guaranteed. Projections and accuracy percentages should only provide perspective and not be solely relied upon to make business decisions. 

In the Forecast accuracy over time table at the bottom, you can compare the Closed won total to the AI projections. An updated AI forecast is provided on day 7, 14, and 21 of every month.

• Closed won: the sum of amounts for all deals with the deal stage set to Closed won.
• Forecast submission: the average forecast submission for the month. If a team submission is available, it is used in the calculation. Otherwise, the sum of rep submission averages is used. Then, the sum is compared to the total sum Amount of deals Closed won each month.
• AI forecast average: the average forecast accuracies for each day projection.
• 7-day projection : the AI projection provided on day 7 of the month.
• 14-day projection: the AI projection provided on day 14 of the month.
• 21-day projection : the AI projection provided on day 21 of the month.

Forecast accuracy for each day projection is measured by computing the absolute percentage difference between the forecasted and actual sales numbers. Average accuracy for a month is measured by calculating the average forecast accuracies of the three projections provided in a month (i.e., 7-day, 14-day, and 21-day projections).

Forecast accuracy is calculated using the following equations:

• Max = (0, 1 - forecast error percentage).
• Forecast error percentage = (|Actual - Forecast| / Actual) x 100.
• When the forecast error is greater than 1, accuracy will always be 0.

Forecast accuracy is dependent on your data being accurate, correct, and up to date.