Time to Conversion

Author: Robin Jia，Tommy Lin

Last Update：2024/06/14

 

1. Model Introduction

The Time to Conversion model focuses on analyzing the time distribution characteristics of users from their first exposure to ads, first page view, first add to cart, first search for specific keywords, and other behaviors to the final purchase completion. It helps understand users' purchase decision cycles and is applied to more reasonably plan and adjust media plans and advertising budgets in the future. It supports constructing custom models by specifying conditions such as time range, ad campaign scope, product scope, keywords, etc.

Through this model, the following business questions can be analyzed:

• After seeing an ad, how long do users generally take to complete an order and purchase? What is the timeliness of the ad?
• After browsing or adding a specific product to the cart, how long do users usually take to complete the purchase? What are the differences in conversion cycles between different products?
• How long after searching for a specific keyword do users complete the purchase?

 

2. Model Interpretation

The main body of the model is a time trend chart reflecting the distribution of cumulative conversion population, intuitively showing the change in the cumulative percentage of users completing conversion as time progresses, which is convenient for judging the overall conversion timeliness. The chart type is a bar chart + line chart, where:

• The bar chart shows the single-period conversion population in each time period, focusing on observing the conversion peak in the short term
• The line chart shows the cumulative conversion population percentage as of each period, which can be used to judge the overall distribution trend accumulated over the long term

This model mainly consists of four pages, which can respectively view the time distribution from the first ad reach to purchase, from detail page view to purchase, from add to cart to purchase, and from keyword search to purchase. Among them, detail page view to purchase and keyword search to purchase can only be viewed using custom models.

In the model's top menu, the predefined model is for fixed Amazon users, all Amazon ad types, and all ASINs under the store. You can select time by month, but the time range is fixed at 3 months. In addition, you can also select the page data you want to view.

The card on the left side of the chart page at the bottom shows some comprehensive data, including the time for 90% of users to complete the purchase, the average purchase time for users, and the time with the most purchases. The bar and line chart on the right side has the horizontal axis as the time interval range from the first ad reach/detail page view/add to cart/keyword search behavior to the purchase. The bar chart represents the number of users who made a purchase within the current time range, and the line chart represents the cumulative percentage of users who made a purchase up to the current time range out of all users. The download button in the upper right corner can download the data table of the model.

 

3. Model Customization

This model supports Amazon Purchase Data Insights and can further customize the four different triggering behaviors.

• First ad reach supports defining ad campaigns, allowing cross-ad type selection of ad campaigns.
• Detail page view and add to cart support defining ASINs. After setting, only the conversion data of the set ASINs will be counted.
• Keywords support entering specific keywords. If multiple keywords are added, users who search for any of the keywords will be counted.

Conversion time in custom models supports custom time grouping. After selecting a fixed interval, the selected time range will be automatically divided into equal numbers of days on average. Using this method, the maximum value of the interval is half of the set time range. Selecting custom event grouping allows customizing non-fixed interval day statistics. Using this method, the maximum value of the interval is the set time range - 1 day.

In addition, because the data of this model needs to reach a certain length of time period, the selected time range cannot be less than 30 days.

 

4. Model Data

Combined with the model data, the distribution pattern of the conversion peak period can be analyzed, as well as how long it takes to cover most of the target conversion population. Based on this, the rhythm and frequency arrangement of ad delivery can be evaluated to determine whether it is reasonable. For example, if most conversions can be completed in the short term, but the lookback window period of the ad audience is too long, there may be budget waste, and the lookback period needs to be compressed; conversely, if the conversion peak period is late, but ads stop being delivered to high-intent users in advance, some conversion populations may be missed, and the rhythm needs to be extended.

Considering that products with longer conversion cycles (such as large home appliances, luxury goods, etc.) may have a right-censoring problem (the observation period is insufficient to cover the complete conversion cycle), when setting the analysis time range, it is recommended to give 1-2 times the conversion cycle as the observation period based on the product characteristics to avoid underestimating long-cycle conversions.

Glossary

Type	Term	Description
Dimension	Traffic event type	Starting event of time to conversion.
	Time Slot	Time Slot
Metrics	Users Purchased	Number of purchase user within time slot
	Conversion Percentage of Purchasing Users	Cumulative percentage of user