How are switch periods chosen in CPP A/B Testing

In Switch tests, only one variant is live at a time. CPP A/B Testing chooses switching periods to help each variant receive exposure under comparable traffic conditions.

This article explains how the system selects hourly, daily, or weekly switching based on minimum traffic thresholds and a fluctuation check. If you want to understand the Switch method first, learn more about the Switch method.

What the system is trying to do

Traffic levels can change by day and by week. To keep exposure fair across variants, the system looks at:

• Whether the ad group has enough traffic for hourly or daily switching, and
• Whether traffic is stable or fluctuating based on the last 28 days of data

Based on this, the system determines switching periods and test duration rules.

Step 1: Check minimum traffic thresholds

To use hourly or daily switching, the ad group must meet minimum traffic thresholds.

Minimum traffic thresholds:

If these thresholds are not met, the system defaults to a weekly switch.

Note: Test duration is calculated based on the traffic and fluctuations of the original ad group.

Step 2: Run the fluctuation check (last 28 days)

The system performs a Fluctuation Check based on the last 28 days of data. The goal is to classify traffic as:

• Stable
• Normally fluctuating
• Highly fluctuating

Fluctuation check steps:

Step 2.1: Calculate weekly average traffic

For each of the last four weeks, the system calculates the average daily traffic.

Where:

• Td,w = daily taps on day d of week w
• Tw = average daily traffic for week w

This creates a baseline for that week.

Step 2.2: Calculate daily fluctuation

For each day, the system calculates how far that day’s traffic is from the weekly average.

Where:

• Dd,w = fluctuation value for day d, week w

The system uses:

• The absolute difference from the weekly average, and
• Division by the weekly average to normalize results

Step 2.3: Sum up weekly fluctuation

For each week, the system sums daily fluctuation values to get a weekly total.

Where:

• Sw = total fluctuation score for week w

A higher Sw value means stronger variation in traffic.

Step 2.4: Determine if traffic has fluctuation

The system checks the weekly totals:

• If Sw ≤ 0.7 for all four weeks, traffic is stable (no fluctuation).
• If at least one week has Sw > 0.7, the system continues to the next step.

Step 2.5: Identify normal vs high fluctuation

For the weeks where Sw > 0.7, the system checks daily fluctuation values:

• If any day exceeds +0.1 for three or more weeks, that day is marked as a high-traffic day.

Then the system classifies the pattern:

• If high-traffic days exist → Normal fluctuation
• If none exist → High fluctuation

Step 3: Apply test duration rules based on fluctuation type

Once the fluctuation type is identified, the system determines test duration rules to ensure each variant gets enough exposure under the traffic pattern.

Duration rules:

Related links

• What is the Switch method in CPP A/B Testing
• Ad Switch vs Ad Group Switch in CPP A/B Testing
• How is test duration calculated in CPP A/B Testing
• Create an Ad Group Switch test for CPP A/B Testing
• Create an Ad Switch test for CPP A/B Testing

Need more help?

If you have further questions on the process, contact your dedicated Customer Success Manager or contact the support team via live chat!