Agents consume activities for each action they perform. If your agent uses more activities than expected, or your quota depletes rapidly, understanding how activities are counted can help you identify the issue and optimize usage.
Symptom
Your agent consumes more activities than you anticipated. You may notice your monthly activity quota depleting faster than expected, receive an email warning that you are approaching your limit, or find that your entire quota was consumed by a single agent or a small number of runs.
Causes
- Multiple actions per run: Each action your agent takes counts as a separate activity. A single run can use many activities if the agent performs multiple steps.
- Processing multiple items: When a trigger returns multiple records (for example, 10 new leads), the agent performs actions on each one, multiplying your activity count.
- Knowledge source lookups: Each search within a knowledge source uses one activity.
- Web browsing and searches: Each web page visit or search query counts as a separate activity.
- Testing on Free plans: Unlike paid plans, tests count toward your monthly quota on Free plans.
- Processing all data instead of new data: Your agent processes all existing records in a data source instead of only new or updated records.
- Loops between agents and Zaps: An agent triggers a Zap that updates data the agent monitors, creating a loop.
How to fix it
Review your agent's activity log:
- From your agent's page, click Activity at the top.
- Expand individual runs to see exactly which actions consumed activities.
- Identify which actions are using the most activities.
Optimize your agent's instructions:
- Keep agents focused on 3-5 related actions rather than creating one agent that does everything.
- Write clear, specific instructions so the agent does not perform unnecessary steps.
- Use agent-to-agent calls to create modular workflows instead of one complex agent.
Process only new data:
When setting up triggers, make sure your agent processes only new or updated records, not all existing records. For example, if you connect a spreadsheet as a data source, configure the trigger to act on new rows only rather than processing the entire sheet.
Avoid loops between agents and Zaps:
If an agent triggers a Zap that updates data the agent monitors, you can create an infinite loop. To prevent this:
- Use filters to stop the loop after one cycle.
- Add a field (like a "processed" checkbox) that the Zap sets after processing, and have the agent ignore records with that flag.
- Use separate data sources for agent input and Zap output.
Use per-run limits as a safeguard:
Each agent run has a built-in activity limit (10 on Free plans, 40 on paid plans). If your agent reaches this limit, it pauses and asks for your input before continuing. This prevents a single run from consuming your entire monthly quota.
If your agent regularly hits this limit, review your instructions to ensure the agent is not trying to do too much in a single run.
Monitor usage regularly:
Check the Activities page to see how your agents are consuming activities. If you notice unexpected spikes, review the agent's activity log to identify which actions are using the most resources.
Learn more about how Zapier Agents usage is measured and best practices for working with Zapier Agents.
Still need help?
If this article did not fix the issue:
- Ask in the Zapier Community. Get advice from experienced Zapier users.
- Learn more about getting help and support with Zapier.