Path to Conversion

Author: Robin Jia，Tommy Lin

Last Update：2024/06/14

 

1. Model Introduction

The Path to Conversion model focuses on the overall process of users' ad interaction before purchase conversion, showing the order of various ad reaches from the first ad reach to the final purchase and their conversion effects. By analyzing the user scale and multi-dimensional conversion metrics covered by different ad sequences, the conversion efficiency of various combination paths can be comprehensively evaluated to find the key ad combinations that have the greatest impact on user decisions and optimize ad delivery combinations and timing strategies. This model supports custom parameters such as analysis time range, ad campaigns, ASINs, etc., and can be flexibly customized for different marketing scenarios.

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

• Before making a purchase, how many ads in total have reached customers and in what order?
• In the past month, what was the subsequent behavior performance of all users who were first reached by a specific ad?

 

2. Model Interpretation

The main body of the model is a user conversion path summary table based on different ad reach sequences.

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 a time range by month.

The table horizontally displays the data of users who have gone through a specific ad reach path, including the total number of unique users, conversion volume, and other indicators. The table vertically breaks down all possible ad reach path combinations. By comparing the performance of various indicators under different sequence combinations, you can intuitively sense the effect differences of different ad combination strategies. In addition, the table also supports sorting by key conversion metrics such as conversion rate, ROAS, etc., which helps to quickly locate the best or worst performing ad reach paths and find out the key influencing factors for improving the overall effect.

Click the edit button in the upper left of the table to open the sidebar. Here you can select more metrics, up to a maximum of 10 metrics displayed simultaneously.

The filtering condition section allows further filtering of conversion paths or metrics and excluding or aggregating data.

• Conversion paths can be filtered in the following four ways, and support combined filtering with multiple conditions (AND/OR)
  • Include: All paths include the specified ad reach
  • Exclude: All paths exclude the specified ad reach
  • Begins With: All paths start with the specified ad for the first reach
  • Ends With: All paths end with the specified ad for the last reach
• Metrics can select a metric currently displayed in the table and filter by value range. It also supports combined filtering with multiple conditions (AND/OR).

The order of ad reach before and after may have a certain impact on conversion effects. To facilitate the analysis of the effects of specific ads in different positions, the table supports filtering paths by conditions such as first touch ad, last touch ad, etc. For example, filtering paths that "end with ad A" can examine the overall performance when ad A is the last reach.

It is worth noting that since the number of unique visitors of conversion paths is negatively correlated with the frequency of ad reach, the sample size of overly lengthy reach paths may be small, and the conversion data may fluctuate greatly, which needs to be appropriately considered in the analysis. The data filtering function can be used to exclude some conversion paths with small data volumes.

 

3. Model Customization

The custom model conditions of this model provide the function of filtering by store/advertiser/ad campaign and support customizing the monitored ASINs. After setting, only the conversion data of the set ASINs will be counted. The ad section below can customize the path nodes. By default, it uses ad types (DSP, SP, SB, SD) as path nodes. Custom nodes can also be selected, with a maximum of 10 different nodes defined.

Click the edit icon to select the ad campaigns included in the node. Note that a single node can only include ad campaigns under the same ad type.

By customizing ad nodes, this model can not only see the conversion paths between ad types but also more finely see the conversion paths between custom ad nodes (product lines/strategies/creative forms, etc.).

 

4. Model Data

The Path to Conversion Analysis model analyzes user behavior based on the order in which users are reached by different ads. It mainly utilizes ad impression, click, and user behavior data such as detail page views and add to carts to construct complete paths from various ad reach to final purchase. The triggering conditions for various ads and user behaviors in the attribution logic are different. When analyzing data, it needs to be noted:

1. DSP and SD-vcpm ads use ad impressions as the triggering condition, while SP, SD-cpc, and SB ads use ad clicks as the trigger. Therefore, DSP ads can often cover more "early reach" audiences with a relatively larger overall scale, while the three types of search ads focus more on "high conversion intent" audiences with a relatively smaller coverage scale.
2. Since the number of DSP ad impressions is far greater than the number of clicks of the other three types of ads, when the conversion path involves both DSP ads and search ads, the main influencing factors of the conversion path may come more from DSP ads. When considering how to optimize the ad combination, it is necessary to fully consider the difference in impression volume.
3. For the purchase of a specific ASIN, in addition to the contribution of various ad channels, the organic conversion of the ASIN (Organic conversion) is also an important influencing factor. The contribution of organic traffic is closely related to the traffic scale of the ASIN itself (natural search ranking, favorites, recommendations, etc.) and may also be indirectly influenced by historical ad delivery. Therefore, in the analysis, it is necessary to have a global control of the overall traffic structure to avoid overly simplifying the interpretation of the ad conversion path.

For the same type of ads that appear continuously in the ad path, the model will perform merge processing during statistics. For example, if a user's ad path is DSP->DSP->DSP->SP (three DSP impressions followed by one SP click), the model will simplify it to DSP->SP. When calculating the path ad spend, the costs of the three DSP impressions will all be counted.

If the user makes a purchase in the path, the path will be split. For example, if a user's path is DSP->SP->Purchase->SB->SP, that is, the user made a purchase after being exposed to DSP and SP ads, and continued to be exposed to SB and SP after the purchase, then this user's path will be split into DSP->SP and SB->SP.

In general, when comprehensively utilizing conversion path analysis to formulate ad optimization strategies, on one hand, it is necessary to fully utilize various behavior data to comprehensively evaluate conversion efficiency; on the other hand, it is also necessary to consider the differentiated positioning of different ad channels in the marketing funnel, not only focusing on the "deep conversion" channels but also taking into account the "broad reach" channels, in order to maximize the synergistic effect.

Glossary

Type	Term	Description
Dimension	Path	Sequence of user engaged ads.
	Ads Account	Advertiser of DSP or entity of Sponsored Ads.
Metrics	UV	Unique Viewer
	Path Size	Occurrence of path.
	Path UV	Deduplicated number of users who have a certain path.
	Total Cost	Total Cost of ads
	Total NTB Product Sales	Total New-to-Brand Product Sales
	Total NTB Purchase	Total New-to-Brand Purchase
	Total NTB Purchase UV	Deduplicated number of new-to-brand purchase users
	Total Product Sales	Total Product Sales
	Total Purchase	Total Purchase
	Total Purchase UV	Deduplicated number of purchase users
	Total ATV	Total Average Transaction Value (Total Product Sales/Total Purchase UV)
	Total CPP	Total Cost per Purchase (Total Cost/Total Purchase)
	Total CPP UV	Total Cost per Unique Viewer Purchase (Total Cost/Total Purchase UV)
	Total NTB CPP	Total Cost per New-to-Brand Purchase (Total Cost/Total NTB Purchase)
	Total NTB CPP UV	Total Cost per Unique Viewer New-to-Brand Purchase (Total Cost/Total NTB Purchase UV)
	Total NTB PR UV	Total Unique Viewer New-to-Brand Purchase Rate (Total NTB Purchase UV/Reach UV)
	Total NTB ROAS	Total New-to-Brand Return on Ad Spend (Total NTB Product Sales/Total Cost)
	Total PR UV	Total Unique Viewer Purchase Rate (Total Purchase UV/Reach UV)
	Total ROAS	Total Return on Ad Spend (Total Product Sales/Total Cost)
	Percentage of Total New-to-brand Purchase	Percentage of Total New-to-Brand Purchase (Total NTB Purchase/Total Purchase)
	Percentage of Total New-to-brand Purchase UV	Percentage of Total New-to-Brand Purchase Unique Viewer (Total NTB Purchase UV/Total Purchase UV)