Path reports

Use path reports to gather insight about the path a given user took that led to a conversion.

Counting conversions

When counting conversions, path reports behave differently than standard reports. Attribution models, settings, or lookback windows aren't taken into consideration and conversions are based only on Floodlight events. Floodlight events are counted as conversions even if they weren't preceded by an impression or click.

Full path

Full path reports can show the number of times a conversion path occurred, and the various events within the path.

Metrics

• Total Paths
  The total number of times a path occurred for unique cookies with a given path pattern.

Dimensions

• Path Pattern ID
  Events with the same path pattern ID are in the same path. This ID is not unique across reports or within multiple runs of the same report.

• Path Event Index
  The position of the event within a given path. Position zero is the earliest.

• Event Type
  The event type within a path. Examples include: View, Click, Floodlight, and Custom.

Path attribution

Path attribution reports shows a count of unique cookies with conversions. You can use custom channel groupings to discover how often given criteria were involved in the converting path. 

Metrics

• Total Paths
  The total number of times a path occurred for unique cookies with a given path pattern.

• Converting Paths
  The total number of unique cookies with one or more last interaction conversions. 

• Path Conversion Rate
  Converting paths / total paths for the entity.

• Total Exposures
  Total impressions + clicks for the entity.

Create a custom channel grouping

You can apply a custom label to any path event that matches a set of rules you've written.

1. Next to Custom channel grouping click + create new.
2. For channel name, write a label you want to appear as a column in your report.
3. For rule names, write labels that will be applied to the events that match your rules.
   1. If an event matches multiple rules, the first rule name it matches will be applied.
4. Add rules using and/or statements.
5. For fallback, write a label that will be shown if none of the rules are met.

Path filters

• You can filter reports to only show the paths containing your chosen information.
• You can also filter by position within a path: Any, First, or Last.

Filter examples

• You might filter your report to only show paths associated with a campaign name that contains the word "video". If you also chose First position the above example would be filtered even more to only show paths with an impression/click in the first position (path event index = 0).

• You might filter your report to only show paths for your most valuable customers, like when a u-variable equals luxury. To do this click Path filters > U Value > U Value Index 0 > Position in path: Any > Equals "luxury" to only show paths that included a conversion where uvar0 == “luxury”.

Report examples

Example - creative sequencing

Let's use a full path report to explore if certain sequences of creatives A and B led to better conversion rates. This will help answer questions like:

• Is Creative A effective on its own or does Creative B help with the conversion?
• Is Creative A effective at all and would we be better off just showing Creative B?

Setup

Download the sample report for creative sequencing (CSV format) to follow along with the explanation below.

Path Pattern ID 0: Creative A → Click → Conversion

157 people saw Creative A, clicked on it, and converted. This resulted in revenue of $1,000.

Path Pattern ID 1: Creative A → Creative B → Click → Conversion

260 people saw Creative A and then Creative B, clicked on Creative B, and converted. This resulted in revenue of $1,900 or an average revenue per conversion of $7.31. ($1900 / 260 = $7.31).

Path Pattern ID 2: Creative A

1800 people saw Creative A and did not click or convert.

Path Pattern ID 3: Creative A → Creative B

2100 people saw Creative A and Creative B and did not click or convert.

Path Pattern ID 4: Creative B → Click → Conversion

790 people saw Creative B and then clicked and converted. This resulted in revenue of $29,000 or an average revenue per conversion of $36.70. ($29000 / 790 = $36.70).

Path Pattern ID 5: Creative B

1,000 people saw Creative B and did not click or convert.

Analysis

• Combining results from path 0 and Path 2, the conversion rate for total people seeing Creative A is 8%. Formula: 157 / (157 + 1800) = 8%
• Combining results from path 1 and path 3,  the conversion rate for total people seeing Creative A and then Creative B is 11%.  Formula: 260 / (260 + 2100) = 11%
• Combining results from path 4 and 5,  the conversion rate for people seeing Creative B is 44%. Formula:  790 / (790 + 1000) = 44%

Take action

Based on our analysis it might be best to focus on Creative B on its own in future campaigns because:  

• Creative B’s conversion rate is quite high when viewed on its own but tends to be lower with Creative A before it (44% vs 11%).
• Paths with Creative A tend to also take more interactions, on average, before a conversion vs paths with only Creative B. 
• Conversions from people who only viewed Creative B were substantially more valuable ($36 vs $7) when Creative A didn't come before Creative B. 

Example - control vs experiment

Let's use a Custom channel grouping in a Full path report to classify one creative as experiment and the others as control. Then we'll see if experiment ads deliver more conversions than Control ads.

Setup

1. Ensure that your experiment creatives contain the word experiment in their names.
2. Create a channel grouping in your path report called "Control Vs Experiment".
3. Ad a rule called "Experiment" to classify all creatives that contain the word "Experiment" as experiment.
4. Set a fallback called "Control" to classify all other creatives as control.

Download the sample report for control vs experiment (CSV format) to follow along with the explanation below.

Path Pattern ID 0: Control → Control → Control → Conversion

96 users saw three control creatives in a row and then converted, resulting in $1000 in revenue. 

Path Pattern ID 1: Control → Control → Click → Conversion

87 users saw two control creatives, clicked on the second one, and converted. These conversions resulted in $755 in revenue. 

Path Pattern ID 2: Control → Experiment → Click → Conversion

142 users saw a control creative and then the experiment creative and then clicked and converted.  These conversions resulted in $1400 in revenue. 

Path Pattern ID 3: Control → Control → Control

3,772 users saw three control ads in a row and did not convert afterwards. 

Path Pattern ID 4: Control → Experiment

822 people saw a control ad and then an experiment ad and did not convert afterwards.

Analysis

• Combining the path and conversion counts between path 0 and path 3, the conversion rate for people seeing three control ads in a row is 2.5%. Formula: 96 / (96 + 3772) = 2.5%
• Combining the path and conversion counts between path 2 and path 4, the conversion rate for people seeing a control ad and then an experiment ad is 15%. Formula: 142 / (142 + 822) = 15% 

Take action

Based on the analysis above, the experiment ad gives significant uplift vs seeing two or more control ads in a row. Maybe try testing the experiment ad a bit more and then potentially replace the control ads.

Limitations

• Path reports have a maximum of 30 days of data.
• Path pattern ID = 0 may display for all path patterns that don't meet privacy requirements of the report. This row is summed.
• Path attribution reports use the last interaction attribution model, and any attribution settings or lookback windows will be ignored.