How to throttle a flow that contains web hooks

Hello ​@Guides for Brides ,

You can manage the webhook rate by batching requests in your sunset flow. Klaviyo doesn’t have built-in throttling, but you can use a delay before the webhook action or segment profiles into smaller groups to control the flow rate. Another option is using an intermediary server or script to queue and send requests at your desired rate (e.g., AWS Lambda, Zapier, or a custom script). This ensures you stay within the webhook limits while still suppressing inactive profiles efficiently. Let me know if you need help setting this up!

Hello ​@Guides for Brides,

Yes! You can throttle your Klaviyo sunset flow to stay within webhook limits using these methods:

1. Add a Time Delay in the Flow

• Insert a time delay between steps to slow down the webhook execution.
• Example: If you set a 1-minute delay for every 500 contacts, you’ll prevent overloading the system.

2. Use Klaviyo’s Batch Processing with Segments

• Instead of triggering the webhook for all profiles at once, create segments based on engagement and process them in smaller batches.
• Example:
  • Segment 1: Last engaged 6–12 months ago → Process Week 1
  • Segment 2: Last engaged 12+ months ago → Process Week 2

3. Use a Third-Party Queueing System

• Instead of directly calling the webhook, send events to an intermediate system (e.g., AWS Lambda, Zapier, or a custom queue service).
• The queue can then process and send requests at a controlled rate.

4. Klaviyo API Rate Limiting (For Suppression)

• If using the Klaviyo API to suppress profiles, ensure you stay within Klaviyo’s rate limits (10 requests per second).
• If needed, use a script with a retry mechanism to handle API throttling.

Kindly reach out if you need further help.

Best Regards,

I think your approach may depend on your technical expertise, or what you have available to you. If you have the tech resource available, then MANSIR2094’s guidance is great.

If you don’t, one approach I’ve taken to spread the load on external systems when calling webhooks via backdating a Flow is to use a Conditional Split, and use the Random Sample filter. Then add different time delays to spread the calls over time. It’s perhaps not the most efficient approach and you can’t be specific about how you adhere to the rate limits.

How many profiles are we talking here? Another approach is to create a segment and then sample it. Within the flow after the webhook set a profile property that indicates that the webhook took place. Make that property an exception in the main segment, and then sample again.

Hi ​@Guides for Brides ! 

I see you’ve received a few great workarounds above. Another option I will suggest is manually suppressing your Sunset segment first - that way that large number of profiles will be suppressed before your flow is set to live, and you won’t risk breaching web hook limits. Then, anyone who enters that segment moving forward will enter your flow and be suppressed through the webhook.