Direct, field-tested answers to the questions teams ask before they rebuild, retry, or call support.
These stable question pages come from recurring client calls, search queries, and support patterns. Each begins with a self-contained answer, followed by causes, diagnostic steps, an example, caveats, and deeper resources.
A Zap usually does not trigger because it is off or unpublished, the trigger is polling and has not checked yet, the source event does not match the selected trigger, or Zapier has already seen the item.
Zapier creates duplicate records when the same event reaches more than one active workflow, a retry repeats a non-idempotent create action, or a find-before-create check uses an unstable lookup value.
A Make scenario commonly misses its schedule because scheduling is off, the interval or timezone is not what you expect, a previous execution is still running, or the trigger returns no new bundles.
An Airtable automation usually does not run because it is off, its record never newly meets the trigger conditions, the selected view or fields changed, or the automation exceeded a workspace limit.
A Zap that worked before usually stopped because an app connection expired, a field or permission changed, the Zap was paused, or the source app changed its payload.
A Zap loops when its action changes the same record watched by its trigger, or when two workflows update each other without an exclusion rule.
A Zapier Filter usually behaves unexpectedly because the mapped value differs from the visible label, is blank at runtime, has a different data type, or uses AND where OR was intended.
Zapier task use rises when triggers emit more items than expected, filters occur after billable actions, loops multiply actions, or overlapping Zaps process the same event.
A Zapier app connection expires when an OAuth token is revoked, a password or permission changes, the provider enforces reauthorization, or the connected user loses access.
A Make webhook receives no data when the sender uses the wrong URL or method, the scenario is not listening or active, a proxy blocks the request, or the sender never dispatched the event.
A Make scenario processes old data when a polling module starts from an earlier point, queued webhook bundles remain, or the source record still qualifies as unprocessed.
Make operation use spikes when modules emit many bundles, iterators multiply downstream work, searches return broad result sets, or error retries repeat modules.
An Airtable trigger fires repeatedly when an automation updates a watched record, the record leaves and re-enters a view, or multiple automations monitor the same condition.
An Airtable formula errors when a function receives the wrong field type, a referenced value is blank or invalid, parentheses are unbalanced, or date and number text is not converted.
An Airtable view trigger misses records when they were already in the view before activation, briefly enter and leave between checks, or the integration watches a different base or view.
An API returns 401 Unauthorized when credentials are missing, expired, malformed, sent in the wrong place, or issued for a different environment.
A 429 response means the integration exceeded an API rate or concurrency limit. The safe fix is to honor Retry-After, slow or queue requests, and reduce unnecessary calls.
A webhook can arrive twice because the sender retries after a timeout, multiple subscriptions target the same endpoint, or both test and production workflows are active.
An API request times out when the server responds too slowly, the request transfers too much data, network resolution fails, or synchronous processing exceeds the automation platform limit.
A mapped Zapier field is blank when the live trigger omits it, the wrong sample path was selected, a previous step did not run, or the field moved inside an array or nested object.
Automation dates shift by one day when a date-only value is interpreted as midnight in one timezone and converted to another, or when daylight-saving rules are applied inconsistently.
Line items map incorrectly when parallel arrays have different lengths, a destination expects one item instead of a list, or an iterator separates fields that must stay together.
A Google Sheets Zap skips rows when rows are inserted above headers, required trigger columns are blank, multiple rows are pasted faster than polling, or formulas populate after the trigger check.
An AI step returns invalid JSON when the prompt allows prose or Markdown, the model invents keys, output is truncated, or unescaped characters break parsing.
CRM records fail to sync when the lookup key is inconsistent, field permissions block writes, required values are missing, or two-way workflows overwrite each other.