# Lead time insights
AWS Supply Chain provides insights on the lead time deviation for a vendor, product, and destination site level. The vendor lead time deviation insights also includes transportation mode, source locations, and identify lead time deviations at a more granular level. You can incorporate the recommended lead times in your planning cycle for enhanced planning accuracy and to avoid stock out risks.
For example, for supplier S, product P, destination site D, source site S, and transportation mode like Truck, Ship, and so on, the **Miss Frequency** displays the frequency of time the lead time was missed, compared to the planned lead time (that is, contractual lead times) shared in the vendor\$1lead\$1time entity. Therefore, Insights recommends to update the planned lead time for the same vendor, product, and site to avoid future lead time issues.
![\[Vendor lead time deviation\]](http://docs.aws.amazon.com/aws-supply-chain/latest/userguide/images/insights_leadtime_deviation.png)
Choose **Export All Recommendations** to export the vendor lead time recommendations for the ingested product, site, or vendor combinations in a .csv file into your Amazon S3 bucket. Once the export is completed, you will receive an email and notification on the AWS Supply Chain web application with a link to the Amazon S3 bucket where the recommendations are exported.
When values for optional columns *source\$1site\$1id* and *trans\$1mode* in the *vendor\$1lead\$1time* data entity are not available, Insights will use the master records for lead times. However, when transactional data for product, source site, destination site, vendor, and transportation mode is at a more granular level, that is, *inbound\$1order\$1line* and *inbound\$1shipment*, it influences the recommendations and the planned lead time. When there are multiple planned lead time records in the master data file, Insights will use the highest planned lead time for calculation.
# Lead time deviations and recommendations
For every generated lead time insight, you can select a row to view the historical trend on the vendor's performance on delivering products from a given ship location to the destination location.
For all orders that are in progress, you can view the status of the order and anticipate the delivery date. Insights uses a machine learning model trained on historical data spanning 1 to 5 years, a time frame chosen during the watchlist creation process, to provide predicted delivery dates with varying levels of confidence.
The **Historical Orders** graph displays the historical average lead times by month calculated from historical order data based on submitted and delivery dates. The bar graphs represent the current planned lead time value and the recommended lead time for vendors at specific sites for the given products. The actual lead time for future orders will be equal or lower than the recommended lead time 50% of the time.
The **Upcoming Orders** graph displays the future purchase order lead times by day, calculated by viewing the order’s submitted date and delivery dates. The bar graphs represent the current planned lead time value and the recommended lead time for vendors at specific sites for the given products. The actual lead time for future orders will be equal or lower than the recommended lead time 50% of the time.
The **Orders in Progress** table displays detailed information of the current or upcoming purchase orders that are at risk based on the model predictions from the historical data for the given vendor, product, and site. The table displays the granular view of all open orders with details such as order quantity, the expected or planned delivery date available from the order line data, and Insights predicted delivery dates with multiple options categorized as *Estimated - Low* and *Estimated - High*. The *deviation* determines the disparity between the estimated high dates and the actual delivery dates available at the order line level.
**Note**
The x-axis in the Historical Orders chart shows months according to the UTC timezone regardless of your location. This means that the beginning of the month coincides with 00h:00m:00s UTC of the first day of the month and the end of the month coincides with 23h:59m:59s UTC of the last day of the month.