When should you pause a campaign, raise the budget or wait? With incomplete data and attribution delays, the answer is rarely obvious. Decision rules written beforehand, with thresholds and deadlines, avoid emotional reactions and make choices consistent and explainable.
The essentials in a definition
Decision rules are criteria written in advance about when to pause, increase or wait on a campaign, even with incomplete data. They avoid emotional, inconsistent decisions and work well with smart bidding, which needs stability. They align with the risk management recommended by NIST: define thresholds, measure and review.
Why have written rules
Without rules, decisions depend on the mood of the day, client pressure or the last report seen. With rules, two different people make the same decision when facing the same data.
They also help explain decisions to third parties and learn from mistakes: if the rule failed, you adjust the rule.
Examples of useful rules
Pause an ad when it has spent X without conversions and its click-through rate is below Y. Raise the budget by 20% when cost per result has been below target for two weeks and the campaign is budget-limited. Wait when volume is below the minimum needed to draw conclusions.
Each rule should state the data, the threshold, the period and the action.
Building a decision rule
- Choose the data and the threshold
- Define the minimum period and volume
- State the action to take
- Record the outcome and review the rule
Dealing with incomplete data
Set a minimum period and a minimum volume before any decision. Allow for attribution delays: conversions from the last few days may not yet be recorded. Prefer comparing equivalent periods and trends, not daily values.
When data is insufficient, the rule should be “wait and review in X days”, not “decide now”.
A good rule decides for us when we are too close to the numbers.
Reviewing the rules
Record each decision and its outcome. Every quarter, review the rules: some will be too cautious, others too aggressive. Adjust based on experience and the business context.
Avoid overly complicated rules. A few clear rules work better than many ambiguous ones.
Decision table
| Situation | Rule | Condition |
|---|---|---|
| Campaign without conversions | Wait until X clicks or Y days | Before pausing, check tracking |
| Cost per customer above limit | Reduce 20% and review after 14 days | Only if data are stable |
| Performance above target | Increase budget 15% per cycle | If stock and capacity exist |
| Sudden drop | Check tracking and changes before acting | Avoid reacting to noise |
Common mistakes to avoid
- Pausing on one day’s data. Daily swings are normal, especially with few conversions.
- Rules without documented exceptions. Define when a rule may be ignored and who decides.
- Not reviewing rules. The business changes; review thresholds each quarter.
How to measure the result
Record each decision, the rule applied and the result after two weeks. If a rule fails several times, adjust it; if it is systematically ignored, review it.
Limits and exceptions
Decision rules do not cover all situations and can be too rigid when context changes (promotions, technical failures, new competitors). They should have documented exceptions and a person who decides when to ignore them. Review them regularly and learn from cases where they failed, instead of treating them as dogma. A rule nobody understands does not help.
Illustrative example
Illustrative example: a team sets that no campaign is paused before 100 clicks and 14 days, unless tracking is wrong. It also sets a maximum 15% budget rise per week. In three months, it cuts reactive changes by two thirds and stability improves. The decision log shows which rules work best.
Frequently asked questions
How many rules do I need?
A few clear ones for the most frequent decisions.
Do rules replace judgement?
No. They are a baseline and can have justified, recorded exceptions.
How do I handle low data?
By setting a minimum volume and waiting before deciding.
How many rules is too many?
Few and clear, covering the most frequent decisions; too many rules nobody follows.
How do I decide with incomplete data?
Define minimum data, deadlines and tracking checks, and use small tests to reduce uncertainty.
Written decision rules (minimum data, deadlines, increase limits and tracking checks) reduce reactions to noise and give consistency to campaign management.
Sources and recommended reading
- Google Ads Help: Smart Bidding
- Google Ads Help: conversion measurement
- NIST: AI Risk Management Framework
Last editorial review: October 2026. Informational content; always confirm current conditions in the official sources before deciding.