Contents
At a glance
- Category
- Automation mechanics
- Polling latency
- Averages half the poll interval, worst case the full interval
- Webhook latency
- Effectively immediate
- Polling cost
- Every scheduled run is billed whether or not anything changed
- Reliability edge
- Polling - a failed poll self-heals next run; a missed webhook is silently lost
- Production pattern
- Webhook for speed plus a low-frequency reconciliation poll for safety
The two ways to find out something happened
Every automation starts with knowing an event occurred, and there are only two mechanisms. Polling is pull: your workflow runs on a schedule, calls the other system's API, and asks what has changed since last time. Webhooks are push: the other system calls a URL you published the instant the event happens - see webhook.
The choice is often made for you by what the source system supports, but where both exist it has real consequences for latency, cost and how the thing fails.
The trade-offs, honestly
Latency. Webhooks are effectively instant. Polling averages half the interval - a five-minute poll means an average delay of two and a half minutes and a worst case of five. For anything where response time is the product, that gap is the whole ballgame; a lead that waits five minutes is measurably less likely to convert than one that waits thirty seconds, which is the point of speed to lead.
Cost. Polling burns your platform's billing unit whether or not anything happened. A five-minute poll is 8,640 runs a month; if two hundred of them find something, the other 8,440 were paid for nothing - which on a per-execution platform is the single most expensive habit in automation. It also consumes the source system's rate limit.
Reliability. Here polling wins, and it is the reason it survives. A poll that fails simply picks everything up next time. A missed webhook is gone unless the sender retries, and you will not know it went missing. Polling is self-healing; webhooks are fast.
What production systems actually do
Both. Webhooks carry the real-time path, and a low-frequency reconciliation poll - hourly, or overnight - sweeps for anything the webhook missed. The sweep asks the source for records created or modified since the last successful sync and processes anything with no counterpart on your side. Because the webhook handler is idempotent, reprocessing an item that did arrive is harmless.
That belt-and-braces pattern costs one extra scheduled run per hour and removes the entire category of "we did not find out for three days". It is the design most worth copying from larger systems into small-business automations.
What you should know
When you must poll, poll intelligently. Always filter by a modified-since timestamp rather than fetching everything and comparing, store the high-water mark from the last successful run rather than assuming the schedule ran, use the API's pagination rather than requesting a huge page, and match the interval to the actual business need - a nightly job does not need to run every five minutes.
And check whether the source offers webhooks before assuming it does not. Many systems have them buried under developer settings, or offer them only on a higher plan, in which case the upgrade may be cheaper than the polling runs it replaces.
Polling vs webhook FAQ
What is the difference between polling and a webhook?
Polling means your automation asks the other system on a schedule whether anything changed. A webhook means the other system calls your URL the moment something does. Polling is pull and self-healing; webhooks are push and instant.
Are webhooks always better than polling?
Faster and cheaper, yes. More reliable, no. A missed webhook is gone silently, whereas a failed poll picks everything up on the next run. The best answer for anything that matters is a webhook for speed plus an occasional reconciliation poll as a safety net.
How much does polling cost me?
A five-minute poll is roughly 8,640 runs a month, every one of which consumes your platform's billing unit whether or not it found anything. If a couple of hundred find work, you are paying for thousands of empty checks - and burning the source system's rate limit at the same time.
How often should I poll?
As rarely as the business genuinely needs. Ask what the cost of a delay actually is: a nightly reconciliation does not need a five-minute interval. Where minutes matter, that is the signal to find a webhook rather than to poll harder.
How do I poll efficiently?
Filter by a modified-since timestamp instead of fetching everything, store the high-water mark from the last successful run rather than assuming the schedule fired, use pagination rather than one huge request, and handle the case where more records changed than one run can process.
What if the system I need has no webhooks?
Check thoroughly first - they are often hidden under developer settings or gated behind a higher plan, and the upgrade can be cheaper than the polling runs it saves. If there genuinely are none, poll on the longest interval the business can tolerate and filter by modified date.
Primary source
n8n - webhook node documentation
Checked August 2026. Anything priced or rate-limited moves, so verify against the source before you budget.
Want this built rather than explained?
Knowing what polling vs webhook means is one thing; having it running in your business is another. We scope one workflow at a time, quote a fixed price in writing after a free consult, build it, and hand it over with documentation you own.
