What are Switch and Parallel methods on CPP A/B Testing

The journey begins in the left‑hand navigation. Open CPP A/B Testing and press Create New. Give the experiment a descriptive name; this name will appear in every table, log, and alert that follows. Here we need to take a pause and give you details on the test methods and types you can utilize. 

Before selecting anything in the creation screen, it is helpful to understand the two test methods and their two implementation types. Review the options below, decide which one fits your strategy, and then make your choice when the platform prompts you.

What is the Switch test method?

A switch test shows one custom product page at a time to your audience, then rotates to another using a switching mechanism that ensures each variant receives equal exposure. The rotation occurs within specified timeframes (hourly, daily, or weekly), allowing all variants to experience traffic under comparable timeframes.

Switching comes in two flavours:

• Switching ad groups
  The platform clones your chosen ad group for every custom product page variant (or the default product page). Only one clone is live at any moment; the others remain paused until their turn arrives.
• Switching ads
  Everything is contained within a single ad group. The tool creates one ad per clone of a custom product page variant and toggles the ad status to provide each variant with equal exposure. A duplicate of the default product page can also be included in this test method. 

How switch periods and test duration are calculated

The content below is more technical and is intended for users who want a deeper understanding of how traffic patterns and fluctuations affect test setup.

If you’re mainly focused on creating or monitoring tests, you can run Switch tests without reviewing these details. The platform applies these calculations automatically.

Minimum traffic requirements for switch periods

To produce meaningful results, the Switch method requires a minimum level of traffic. The required threshold depends on the selected switch period.

• For an hourly switch, the ad group must receive at least 100 taps or 30 installs per 6–8 hour slot, depending on the number of variants in the test.
• For a daily switch, the ad group must receive at least 100 taps or 30 installs per day.

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

Test duration is calculated based on both the traffic volume and the fluctuation behavior of the original ad group.

Fluctuation check 

Traffic levels can vary across days and weeks. To provide fair exposure for each variant, CPP A/B Testing runs a fluctuation check using data from the last 28 days.

This check determines whether the selected ad group’s traffic is:

• Stable
• Normally fluctuating
• Highly fluctuating
  
  

The result of this check is used to select the appropriate switch period and calculate test duration.

Calculating weekly average traffic

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

Where:

• T_d,w = Daily taps on day d of week w
• T_w = Average daily traffic for week w

This establishes the baseline activity level for each week.

Calculating daily fluctuation 

For each day, calculates the deviation from the weekly average:

 

Where:

• D_d,w = Fluctuation value for day d, week w
• The absolute difference measures how far a day’s traffic is from the average
• Dividing by the weekly average normalizes the result so it can be compared across weeks

Summing up weekly fluctuation

Where:

• Sw​ = Total fluctuation score for week w
• A higher Sw value indicates stronger variation in traffic

Determining if traffic has fluctuation

Uses the weekly totals to determine if traffic is stable or variable:

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

Identifying normal vs. high fluctuation

Reviews the weeks where S_w > 0.7

For those weeks, checks each day’s fluctuation values:

• If any day exceeds +0.1 for three or more weeks, mark it as a high-traffic day.

Classifies the fluctuation:

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

Test duration rules

Once the fluctuation type is identified, the system determines the test duration to ensure each variant receives enough exposure under both high and normal traffic conditions.

Fluctuation Pattern	Recommended Switch Option	Test Duration Formula
High Fluctuation	Weekly	Variant Count x Week
Normal Fluctuation	Daily/Hourly/Weekly	Variant Count x Week
No Fluctuation	Daily/Hourly	Variant Count x Day

What is the Parallel test method?

A parallel test keeps all variants live simultaneously, so the comparison comes from side‑by‑side traffic rather than variant rotation. This method is recommended for users who want to minimize the impact of seasonality by ensuring that all variants run simultaneously under the same market conditions.

 

​​Parallel also offers two distinct paths:

• One ad group, multiple custom product pages

You pick one ad group and choose up to four custom product pages (the default product page can be one of them). The original ad group is paused, and all duplicate ads run side‑by‑side for the full test window, letting you see which page wins with identical keyword, bid, and audience settings at the same time frame.

• Multiple ad groups, one custom product page

You select up to four existing ad groups and then pick one custom product page (selecting the default is not possible in this case). The system attaches the selected custom product page to each ad group. Every ad group, with the same custom product page, runs concurrently, so you can spot which ad‑group setup (keywords, bids, audiences) delivers the best results for that page.

Let’s dive into how you can create a test with each method and type. 

Click here to learn how to set up CPP A/B tests with the Switch method

Click here to learn how to set up CPP A/B tests with the Parallel method