Forecasting AI

[CDiVincenzo]

Dear Hubspot Community,

I would like to know if someone has used the new tool Forecasting AI.
Theoretically it sounds quite useful, but in my case checking vs the won amount only last month looks more reliable than our submission.
In the previous months our submissions are closer to the final amount than the data from Forecasting AI.
How is your experience? 

[aviafriat26 · Accepted solution]

Hey @CDiVincenzo, 

I've been tinkering with Forecasting AI, and I totally get where you're coming from. It's a nifty tool, but it can be hit or miss depending on your data and how consistent your sales process is.

Here's what I've found:

• Clean, Historical Data: Forecasting AI works best with lots of clean, historical data. If your past submissions are more accurate, it could be that your team just has a great gut feel that the AI hasn’t caught up with yet.
• Use as a Second Opinion: Treat Forecasting AI as a second opinion rather than the ultimate truth. Compare its forecasts with your manual ones over time. You might see the AI get better as it learns your patterns.
• Data Quality: Make sure your data is clean and consistent. Any junk data can throw off the AI.
• Patience: Give it some time. AI models need to learn, and they get better with more data and time.

Hope this helps! Would love to hear more about your experience with it.

[nlafakis · Accepted solution]

I've spent the last 6 months living in the reporting tool and I can tell you hands-down that the AI-forecasting tool needs a lot of love. There are some really simple reports that I've asked it to make that do not require clean data (ex: something as simple as - How many contacts have been created over the last year so far by month?) and it still cannot manage to produce a report worth looking at. 9/10 times it doesn't produce anything at all and tries to use custom equations for simple calculations. I have yet to use a single one and it's so far been much faster to build them than to continue playing with it.

I do look forward to it being more reliable and being able to execute more accurately on simple reports that we create often.

[CDiVincenzo]

Thank you @aviafriat26 @PamCotton 

[aviafriat26 · Accepted solution]

Hey @CDiVincenzo, 

I've been tinkering with Forecasting AI, and I totally get where you're coming from. It's a nifty tool, but it can be hit or miss depending on your data and how consistent your sales process is.

Here's what I've found:

• Clean, Historical Data: Forecasting AI works best with lots of clean, historical data. If your past submissions are more accurate, it could be that your team just has a great gut feel that the AI hasn’t caught up with yet.
• Use as a Second Opinion: Treat Forecasting AI as a second opinion rather than the ultimate truth. Compare its forecasts with your manual ones over time. You might see the AI get better as it learns your patterns.
• Data Quality: Make sure your data is clean and consistent. Any junk data can throw off the AI.
• Patience: Give it some time. AI models need to learn, and they get better with more data and time.

Hope this helps! Would love to hear more about your experience with it.

[PamCotton]

Hey @CDiVincenzo, Thank you for reaching out to the HubS@pot Community!

I understand your concerns regarding the accuracy of Forecasting AI compared to your own submission data. I want to invite our top experts to share their thoughts, @aviafriat26, and @LouiseHare do you have any recommendations for the @CDiVincenzo matter?

Thank you,

Pam