> ## Documentation Index
> Fetch the complete documentation index at: https://docs.metricanic.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Choose a rule schedule

> Balance detection speed and conversion delay by choosing the evaluation window, run frequency, and recent traffic exclusion.

A schedule controls when an automation rule looks for matches and which report data it evaluates. Choose it around the time your traffic needs to produce an outcome.

A lead that arrives several hours after a click can make fresh traffic look unproductive. Excluding recent data reduces that problem, while a longer time frame gives low-volume placements more evidence to accumulate.

## Separate the three decisions

| Setting           | What it controls                       | Tradeoff                                                                                              |
| ----------------- | -------------------------------------- | ----------------------------------------------------------------------------------------------------- |
| **Time frame**    | How much eligible history is evaluated | Short windows react faster. Long windows give a larger sample but retain older performance            |
| **Frequency**     | How often the rule is scheduled        | More frequent checks find changes sooner but do not make underlying data or conversions arrive sooner |
| **Ignore recent** | How far back eligibility ends          | A delay allows conversions to arrive, at the cost of reacting later                                   |

Changing frequency does not turn the rule into a real-time trigger. Saving a rule schedules future evaluation. It does not immediately tag every row that currently matches.

## Start with the conversion delay

For a campaign whose useful leads usually arrive within an hour, a reasonable starting point is a 1-day time frame, a 2-hour frequency, and a 1-hour recent exclusion. Evaluate whether the resulting tags identify useful review candidates, then adjust based on observed delay and traffic volume.

For a new, low-volume campaign, increasing frequency alone gives you repeated checks of the same small sample. A longer time frame and a minimum-volume or minimum-cost condition are usually more useful.

Edit the rule in **Automation Rules**, set **Time frame**, choose an offered **Frequency**, then choose **Ignore recent** and save. The available choices depend on the time frame.

## Understand the hour boundaries

Rules evaluate UTC hourly report buckets. This matters when comparing a rule with a report that uses a different timezone or an exact minute range.

With **Ignore recent** set to zero, the current partial hour is eligible. With a positive exclusion, Metricanic stops before the hour containing the cutoff time. It then counts backward by the selected time frame.

For example, a run at **14:40 UTC**, with a **3-hour** time frame and **30-minute** recent exclusion, has a cutoff at 14:10. The 14:00 bucket is excluded, so the rule evaluates **11:00 through 13:59:59 UTC**.

That result is intentionally different from a minute-by-minute 11:10 to 14:10 window. Match the eligible hour buckets when checking why a row received a tag.

## Available combinations

All values below are minutes except the time-frame labels.

| Time frame | Frequency       | Ignore recent      |
| ---------- | --------------- | ------------------ |
| 1 hour     | 30, 60          | 0, 15, 30          |
| 3 hours    | 30, 60, 120     | 0, 15, 30, 60      |
| 12 hours   | 60, 120, 240    | 0, 30, 60, 120     |
| 1 day      | 120, 240, 480   | 0, 60, 120, 240    |
| 2 days     | 240, 480, 960   | 0, 120, 240, 480   |
| 3 days     | 240, 480, 960   | 0, 120, 240, 480   |
| 7 days     | 480, 960, 1440  | 0, 240, 480, 960   |
| 14 days    | 960, 1440, 2880 | 0, 480, 960, 1440  |
| 30 days    | 1440, 2880      | 0, 960, 1440, 2880 |

The shortest frequency is 30 minutes. A 15-minute recent exclusion is available for some windows, but a 15-minute execution frequency is not.

If a tag arrives later than expected, check that the rule is enabled and that the chosen window contains matching data. The scheduler and report processing are asynchronous, so a frequency is not a promise of execution at an exact wall-clock minute. Existing tags also persist after a row stops matching, as explained in [Create automation rules](/automation/rules).