- Burst and steady: two limits on every call
- The five tiers and other hard limits
- What a 429 tells you, and how to respond
- Why a loop that looks safe still gets blocked
- Who shares the quota
- Event data limits
- How Slack tiers its API for comparison
- My advice for anyone building on Klaviyo
- Frequently asked questions
The Klaviyo API rate limit comes in five tiers, from XS at 1 request per second and 15 per minute up to XL at 350 per second and 3500 per minute. Most endpoints use one of these unless their docs say otherwise. Go over and you get an HTTP 429 with a Retry-After header in seconds.
Burst and steady: two limits on every call
Klaviyo gives each endpoint two caps that work together. The burst limit is per second and stops sudden spikes. The steady limit is per minute and stops a long run of requests from wearing the system down. Hitting either one returns an HTTP 429, so a script that stays under the per-second number can still get blocked if it keeps going for a full minute.
Unless an endpoint's own documentation says otherwise, it sits on one of five tiers. The tier for each endpoint is listed in that endpoint's docs, so check there rather than assuming.
The five tiers and other hard limits
| Item | Limit | Applies to |
|---|---|---|
| XS tier | 1/s burst, 15/m steady | Endpoints on the XS limit |
| S tier | 3/s burst, 60/m steady | Endpoints on the S limit |
| M tier | 10/s burst, 150/m steady | Endpoints on the M limit |
| L tier | 75/s burst, 700/m steady | Endpoints on the L limit |
| XL tier | 350/s burst, 3500/m steady | Endpoints on the XL limit |
| Payload size | 5 MB decompressed | Klaviyo's new APIs |
| Event properties per packet | 400 | /api/events, /client/events, legacy /api/track |
The gap between tiers is huge. An XS endpoint allows 15 calls a minute, while an XL endpoint allows 3500. Plan your sync around the slowest endpoint you need, because that's the one that will decide how long a full run takes.
What a 429 tells you, and how to respond
On a normal response, Klaviyo sends three headers describing the steady window: RateLimit-Limit, RateLimit-Remaining and RateLimit-Reset. Read them and you'll know how close you are before you hit the wall.
When you do hit it, those headers are replaced by a Retry-After header giving the number of seconds before requests can resume. The right response is to pause for exactly that long, then carry on. Hammering the endpoint again straight away just burns more of your window. Rate limits also reset on the same window across the related limits, so one pause covers all of them.
Why a loop that looks safe still gets blocked
Klaviyo Developers documents a trap most people only find in production. Requests that use additional-fields or include query parameters are held to a stricter, lower rate limit than the same call without them. And a single call can consume several limits at once. In Klaviyo's worked example, one call combining include=lists and additional-fields[profile]=predictive_analytics counts as three calls across the limits.
So if you tested your loop against the plain endpoint, then added an include to save a second request, you may have tripled its share of the limit. Work out which parameters each call uses before you set your pacing.
Who shares the quota
OAuth apps get their own quota for each installed app instance, so per account per app. Private key integrations share one quota per account. If you run three private key scripts against the same Klaviyo account, they're all drawing from the same pot, and one busy script can slow the others down. Moving a heavy integration to OAuth gives it its own allowance.
Event data limits
- 400 event properties per data packet.
- 100 kB maximum per field.
- 4000 items per array.
- 11 levels of nested objects.
- Event timestamps between 1990 and one year from now.
These sit alongside the 5 MB decompressed payload limit on the new APIs. If you're sending rich order data, the array and nesting limits are the ones to watch.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
How Slack tiers its API for comparison
Slack's Web API also uses tiers, from Tier 1 at 1+ per minute to Tier 4 at 100+ per minute, applied per method per workspace per app. The idea is the same: different endpoints get different allowances. Mailchimp documents its Marketing API fundamentals separately and doesn't use Klaviyo's tier names, so check its own docs if you're moving between the two. My Klaviyo and Mailchimp comparison covers the wider differences.
My advice for anyone building on Klaviyo
Read the rate headers on every response, honour Retry-After to the second, and count include and additional-fields calls as heavier than they look. If several tools share one private key, give the busiest its own OAuth app. And if you're relying on the API heavily, a good API monitoring tool will show you the 429s before your customers notice. For a broader view of the platform, my Klaviyo review is the place to start.
Want your tool on this list? Featured spots and reviews here are paid partnerships. Here's how to partner with me.
Frequently asked questions
What happens when I exceed the Klaviyo rate limit?
You get an HTTP 429 and a Retry-After header with the seconds to wait.
Is the limit per account or per app?
OAuth apps have their own quota per installed app instance. Private key integrations share a quota per account.
Do all endpoints have the same limit?
No. Each uses one of the XS to XL tiers unless its documentation says otherwise.
How can I see how many requests are left?
Read the RateLimit-Limit, RateLimit-Remaining and RateLimit-Reset headers on non-429 responses.
Are there size limits on events?
Yes: 400 properties per packet, 100 kB per field, 4000 array items and 11 nesting levels.
