Unique Reach

 

1. Model Introduction

The Unique Reach Analysis model focuses on the number of unique audiences covered by different ad campaigns, showing the trend of audience reach and the change in corresponding ad spending. By observing various indicators, the new user reach efficiency of ads can be evaluated. The model supports custom parameters such as time range, ad scope, time granularity, etc., flexibly meeting personalized analysis needs.

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

• What is the overall trend of the number of users reached by various ads?
• What is the proportion of new users reached by ads daily? Is the expansion capability good?
• What is the average ad cost per incremental reached user?

 

2. Model Interpretation

This model is divided into two main parts. The card section at the top shows the summary data related to ad reach, and the chart section at the bottom shows the trend of ad reach-related data per day/week/month.

In the model's top menu, the predefined model is for fixed Amazon users. You can single-select all Amazon ad type data and single-select ad campaign data in DSP ads. To view ad campaign data of other ad types, you can use custom models. See the model customization section for details. The time selector allows you to freely select cross-month data.

The summary metric cards show the key reach metrics within the selected time range, such as total reach, percentage of new reach, and average cost per new reach. You can click the download button in the upper right corner to export the data as an Excel spreadsheet for further analysis.

It's worth noting that the summary metric cards by default show the data for the last month within the selected time range. To view cross-month summary data, you can use custom models for analysis (see the "Model Customization" section).

The trend chart at the bottom by default shows the changing trends of metrics such as ad reach, new reach, percentage of new reach, average cost per new reach, etc. over time. The time granularity can be switched between daily/weekly/monthly.

Click the edit button in the upper left corner to open the sidebar, where you can select more metrics, up to a maximum of 6 metrics displayed simultaneously.

Click the trend chart at the bottom to change the metrics corresponding to the bar chart and line chart in the trend chart. You can select up to four metrics to display.

 

3. Model Customization

The custom model conditions for this model allow you to select a custom time range down to specific dates and support selecting the statistical time interval. The ads section below allows you to select the ads to be counted.

• By ad type
  • Ad reach of Amazon's ad types respectively
• By campaigns/orders
  • Ad reach of each ad campaign under the advertisers or stores in the Instance
• By customized campaign/order group
  • Ad reach statistics in the form of custom groupings, which can be counted by ad type or by ad campaign. If multiple ad campaigns are added to a group, these ad campaigns will be treated as a whole (deduplicated data) to count ad reach. Regardless of the counting method used, the ads in the same group must come from the same ad type.

 

4. Model Data

The statistical logic of the Unique Reach model is as follows:

1. Logic of Deduplication by Number of People

When calculating the number of unique users reached by ads, the method used is to deduplicate and calculate based on the selected ad scope. The specific deduplication calculation method varies depending on different data aggregation forms, which will affect the statistical caliber of the reach count.

For example, when selecting to count data by day, the system will first count the number of people reached by ads each day. If the same user is reached multiple times by the same ad on the same day, that user will only be counted once. But if that user is reached again the next day, they will be counted in the reach of the next day. Among them, only the initial reach day's reach is counted as new users.

When selecting to count data by week or month, user deduplication statistics will be based on the unit of week or month. That is, the same user is only counted once within each week/month.

Similarly, selecting different ad statistical scopes (such as counting independently by ad campaign or aggregating by ad type) will also affect the minimum unit of data deduplication, thereby affecting the statistical reach count.

2. Logic of New Exposed and New Clicked Users

When the user determines the time window and time granularity in the customization process, the data is divided into several segments. For example, if the time window selected is 1.1 - 1.31 and the time granularity is day, then the data will be divided into 31 segments.

When counting the number of new exposed and new clicked users, we will count the users as new exposed or clicked users of that time segment based on the time segment in which each user's first exposure or click appears within the statistical time window.

For example, when counting with a time granularity of days in the time window of 1.1 - 1.31, if a user is exposed to an ad on 1.2, 1.5, and 1.12, then when counting the number of new exposed users on 1.2, that user will be counted. However, that user will not be counted in the number of new exposed users on 1.5 and 1.12. The same applies to counting the number of new clicked users.

It needs to be specially explained that the "new reach users" in the model refers to users who have never been reached by the selected ads in the past within the selected time range. In the predefined model, the new users in the data card are compared to the last month versus the previous six months. The data card of the custom model will not display the summary of new users and their percentage because it lacks a historical period for comparison.

In the trend chart, the calculation of new users is related to the selected time granularity. Taking daily statistics as an example, the number of new users on a certain day refers to users who have not been reached by the selected ads from the previous day (the first day within the user-defined time range for the custom model, and 6 months ago for the predefined model) to that day.

Glossary

Type	Term	Description
Dimension	Time Slot	Time slot of statistics
	Advertiser	Advertiser of DSP or entity of Sponsored Ads.
	Ads	A specific ad product type in DSP/SP/SD/SB or a collection of ad campaigns or a single ad campaign.
Metrics	UV	Unique Viewer
	Total Cost	Total ad spend of ads.
	Total Cost of New Reach UV	Ad cost of new impression user.
	Total Cost of New Click UV	Ad cost of new click user.
	Impressions	Number of impression event.
	Click-throughs	Number of click event.
	Reach UV	Deduplicated number of impression users (Reach Unique Viewer).
	Click-throughs UV	Deduplicated number of click users (Click-through Unique Viewer).
	eCPM	Effective Cost per Thousand Impressions (Total Cost/Impressions*1000)
	eCPC	Effective Cost per Click (Total Cost/Click-throughs)
	New Reach UV	Deduplicated number of new impression users.
	New Click UV	Deduplicated number of new click users.
	Percentage of New Reach UV	Percentage of New Reach Unique Viewers (New Reach UV/Reach UV)
	Percentage of New Click UV	Percentage of New Click Unique Viewers (New Click UV/Click-throughs UV)
	Frequency UV	Frequency per Unique Viewer (Impressions/Reach UV)
	Cost per Reach UV	Cost per Reach Unique Viewer (Total Cost/Reach UV)
	Cost per Click UV	Cost per Click-through Unique Viewer (Total Cost/Click-throughs UV)
	Cost per New Reach UV	Cost per New Reach Unique Viewer (Total Cost/New Reach UV)
	Cost per New Click UV	Cost per New Click Unique Viewer (Total Cost/New Click UV)