Exception Management Frequently Asked Questions
Exception Management (EM) is currently a limited release feature that’s not available to all customers. Eligibility depends on data volume and model training readiness; accounts with a large amount or insufficient historical data may be restricted. If you can’t access Exception Management, contact your NetSuite Account Manager for guidance and next steps.
Additionally, Exception Management is a production‑only feature and can't be enabled in Sandbox Accounts because those environments typically lack the required historical transaction data (approximately 18 months) needed for model training and inference.
Below are some of the most common questions received when working with Exception Management (EM):
General
What is an exception?
Exceptions are transactions in NetSuite that appear to be outliers based on your historical patterns.
Currently, the EM dashboard identifies transactions with unusual transaction amounts and accounts, with an 'Expected Transactions' tab that displays missing transactions.
Who should use Exception Management?
Exception Management is for NetSuite users who work in the financial or accounting teams. This includes anyone who is responsible for ensuring all recorded transactions are accurate and up to date.
How can I enable Exception Management?
To enable Exception Management, as an Administrator, go to Setup > Company > Enable Features > Accounting subtab, then under Advanced Features, select EXCEPTION MANAGEMENT. For more information, see Setting Up Exception Management,
Why isn’t EM available in Sandbox Accounts?
EM requires sufficient historical transaction data to train customer-specific models. Sandbox environments typically don't have enough data (approximately 18 months of stable history) needed for reliable training and inference.
Can I customize the criteria for an exception?
Yes, you can set preferences at Setup > Accounting > Accounting Preferences > Exception Management tab. For more information, see Using the Exception Management Preferences.
Interface
Where can I find my identified exceptions?
Go to Lists > Accounting > Exception Management.
How often does EM detect exceptions?
Every hour, EM reviews all transactions that were created or edited in the past hour. Edited transactions, even if they were previously labeled as exceptions, will be re-evaluated by the model—if there is still an issue with the transaction, a new exception will be created.
What should I do with my exceptions?
Ideally, review each exception and determine if they are properly-entered transactions or not. You can correct any incorrect transactions, or indicate that exceptions are not an issue. For more information, see Resolving Exceptions.
What if EM marked a transaction as an exception, but I know the transaction is correct?
Because EM looks for transactions that are different from your usual patterns, it will flag and alert you of any transaction which is off the normal pattern. When this happens, click on the exception's View Details, then click No Action Needed.
For more information, see Resolving Exceptions.
As you resolve more exceptions, EM will get better at identifying your transaction patterns and will offer more accurate suggestions.
AI Model
What data does Exception Management (EM) use to learn my transaction patterns?
EM uses your transaction data of up to 18 months to understand your transaction patterns. This allows it to make sense of your most recent data trends to make suggestions. Your organization's data is only used to train a dedicated custom model. This model will only make suggestions relevant to your organization.
When training, the models search for recurrence patterns from which predictions are made.
What is the minimum amount of transaction data I need to use Exception Management?
For best results, Exception Management (EM) requires historical transaction data of up to 18 months to learn from. After you have enabled EM, it begins learning from the data in your records. Then, every hour, EM reviews transactions that you created or edited in the past hour.
EM is more accurate if you have consistent data to learn from over the last year. Its performance will be impacted if your business changed recently, or if you have insufficient data for it to learn from.
How does EM learn from my feedback?
EM's machine learning capabilities improve with user feedback.
Dismissing a suggested exception provides feedback to the system, indicating that the transaction is not an exception. Similarly, resolving a suggested exception validates that the suggestion was correct. Therefore, both positive and negative feedback help EM improve the accuracy and quality of the suggestions it can make.
Sometimes, EM may display similar exceptions that were previously dismissed, and user feedback on those exceptions is valuable for the system to improve its suggestions for similar transactions in the future.
Model and feature experience will be evolving over time based on user feedback. Therefore, consistent feedback is valuable to the system.
Is EM available for all transactions?
The feature covers over 30 transaction types, including custom transactions.
Missing transactions cover transaction types that are likely to have recurrence patterns.
Incorrect Amount and Incorrect Account exceptions cover most transactions that have an amount and account property, respectively.
How often is the model retrained?
The models for Incorrect Account and Incorrect Amount are retrained weekly for any recent trends in your data.
The Missing Transaction model is retrained at the beginning of every accounting period, predicting transactions through the period that started and surfacing missing transaction exceptions after the expected date of the transaction has passed.