Sync latency

Nektar’s design strikes a balance between efficiency and accuracy. Nektar is tuned to meet the real-world latency requirements for CRM data, without being wasteful of API quotas or compute resources.

Typical end-to-end latency

This table shows the typical end-to-end latency for reading and processing new data and writing it to Salesforce.

What does “end-to-end latency” mean?

• We start the clock when:

  • an email is sent or received in a Nektar-monitored mailbox
  • an event is added to or edited in a Nektar-monitored calendar
  • the manual Salesforce change is made, for self-healing and auto-recall

• We stop the clock when Nektar makes the corresponding change on Salesforce.

Something appears to have taken longer

This could happen because:

• Nektar sync might have been disabled or stopped due to error for the particular user or object at the time of the start-clock event mentioned above.
• Nektar sync was only recently enabled for this user or object, and the initial bulk sync has not completed. These latencies are applicable to ongoing operations, not during bulk sync. The sync status icon is blue during bulk sync and green during ongoing operations.
• Salesforce writes might have been disabled or stopped due to error at the time.
• Salesforce writes were only recently enabled, and it had not completed bulk sync yet.
• Events (including recurring) more than 3 months in the future are only processed by Nektar when they come within the 3 month window.
• Auto-recall. For example if there was no matching account at the time of an email but the account was created later, the account creation is the start-clock event, not the email itself.
• An event was originally filtered out due to a screening rule, but a subsequent edit made it pass. The edit is the appropriate start-clock event, not the creation.
• Salesforce API errors might prevent Nektar writing data for prolonged periods of time. These often occur due to poorly performing Apex triggers or other tools that perform too many API calls to your Salesforce instance.
• In rare circumstances, Nektar, the communication tool’s API service, or Salesforce might have encountered a technical incident at time of the activity. Nektar’s uptime status for the past 30 days is available, as are similar pages for other tools.

Sources of latency

There are three main sources of latency in Nektar:

1. Nektar is not the tool used by your sellers to send emails, so it must fetch these emails periodically from each inbox and calendar. Fetching too frequently will exhaust Google / Microsoft API quotas, so Nektar strikes a balance at 10–15 minutes.
2. Salesforce API quotas are also quite limited, and it is hundreds of times more efficient to write data to Salesforce in batches than one-at-a-time. To strike a balance, Nektar aggregates changes over 10–15 minutes to write data in reasonably-sized batches.
3. Nektar achieves its unparalleled accuracy with a deep, complex and computationally intensive graph processing pipeline. The infrastructure is sized with ample headroom for ongoing operations, but can occasionally be backed up when performing bulk operations such as historical scans for many new users.

Step-by-step breakdown

• Nektar reads each user’s communication data, and your organization’s Salesforce data, once every 10–15 minutes. Each type of data and each user’s communication data is read independently. (Average delay: 7 minutes; Maximum delay: 15 minutes)
• Nektar processes the data in our pipeline: constructing, de-duplicating and linking the new activities, contacts and contact roles, as well as re-evaluating all related past activities, contacts, contact roles and opportunities. (Average delay: 15 minutes, Maximum delay: 4 hours in congested conditions)
• Intentional wait to ensure that all related records have been independently fetched and unified before writing to Salesforce. This includes manual Salesforce updates made around the time of the activity. (Wait time: 30 minutes before creates and updates, 24 hours before contact merges)

• Nektar writes new data into Salesforce every 10–15 minutes. Each object type is written independently. Note:

  • Activities and Opportunity-Contact Roles can only be created after their Contacts have already been created, and the Salesforce Contact records have been processed by Nektar.
  • A fraction of Salesforce writes fail due to congestion at Salesforce’s end, and need to be re-tried. We retry up to multiple times with increasing delays for up to 4 hours.

  (Average delay: 8 minutes, Maximum delay: 4 hours 15 minutes)