Conversion Attribution Models
Table of Contents
- What is an attribution model?
- Example conversion path
- Linear
- First Touch
- Last Touch
- Positional
- Time Decay
- Full Attribution
- Compare the models
- Campaign reports vs. Flight and Project reports
- Which model should I use?
- How to change the attribution model
- Common Questions
What is an attribution model?
People rarely convert after seeing just one campaign. A supporter might click a display ad, open an email a week later, and then donate. An attribution model is the set of rules Feathr uses to decide which campaigns get credit for that conversion and how the conversion value is divided between them.
Each model answers two questions:
- Which campaigns show the conversion? Some models show the conversion on every campaign the person interacted with. Others show it on only one campaign.
- How much of the conversion value does each campaign get? Some models split the value between campaigns. Others give the full value to each campaign.
A campaign that shows the conversion always counts it as one full conversion. The model changes how much of the conversion value that campaign is credited with.
If a person interacts with only one campaign before converting, that campaign gets the conversion and its full value no matter which model you choose.
Note: A campaign can only receive credit for a conversion if the person completed that campaign's goal after seeing or clicking its ads. Learn more about goals.
Example conversion path
Every model below uses the same example. A prospect is considering registering for your conference:
- They find your website for the first time by clicking an ad from Campaign A.
- A few days later, they come back by clicking an ad from Campaign B, but they don't register.
- Over the next week, they see ads from Campaign C.
- They go directly to your website and register for $100.
Feathr records each of these interactions on the person's record. Together, they make up the person's path to conversion. Feathr uses this path to decide how to credit the $100 registration to Campaigns A, B, and C.
Linear
Every campaign in the path shows the conversion, and the value is split evenly.
In the example, Campaigns A, B, and C each show one conversion. The $100 is divided equally, so each campaign is credited with $33.33.
| Campaign A | Campaign B | Campaign C | |
|---|---|---|---|
| Conversion | 1 | 1 | 1 |
| Conversion value | $33.33 | $33.33 | $33.33 |
Best for: A balanced view of how your campaigns work together. Linear is a good fit when every touchpoint plays a role, like a multi-channel appeal with ads and emails running at the same time.
Keep in mind: Linear treats a campaign the person saw once the same as one they engaged with many times.
First Touch
Only the first campaign in the path shows the conversion, and it gets the full value.
In the example, Campaign A first brought the prospect to your website, so it shows the conversion and the full $100. Campaigns B and C don't show the conversion.
| Campaign A | Campaign B | Campaign C | |
|---|---|---|---|
| Conversion | 1 | — | — |
| Conversion value | $100 | — | — |
Best for: Understanding which campaigns introduce new people to your organization. First Touch is helpful for awareness and prospecting campaigns that find new audiences.
Keep in mind: First Touch ignores the campaigns that kept the person engaged and helped them make a decision.
Last Touch
Only the last campaign before the conversion shows it, and it gets the full value.
In the example, Campaign C was the final campaign the prospect interacted with before registering, so it shows the conversion and the full $100. Campaigns A and B don't show the conversion.
| Campaign A | Campaign B | Campaign C | |
|---|---|---|---|
| Conversion | — | — | 1 |
| Conversion value | — | — | $100 |
Best for: Understanding which campaigns close the deal. Last Touch is helpful for retargeting and deadline campaigns, like "Registration closes Friday."
Keep in mind: Last Touch ignores the campaigns that first introduced the person to your organization.
Positional
Every campaign in the path shows the conversion. The first and last campaigns get most of the value.
The first and last campaigns are each credited with 40% of the conversion value. The remaining 20% is split evenly between every campaign in the middle. In the example, Campaign B is the only campaign in the middle, so it gets the full 20%.
| Campaign A | Campaign B | Campaign C | |
|---|---|---|---|
| Conversion | 1 | 1 | 1 |
| Conversion value | $40 | $20 | $40 |
If there were four campaigns in the path, the two middle campaigns would each get 10%, or $10.
Best for: Giving the most value to the campaigns that introduce people and the ones that convert them, while still recognizing the campaigns in between.
Keep in mind: The middle campaigns share a small portion of the value, even if they did a lot of the work.
Time Decay
Campaigns closer to the conversion get more of the value.
Time Decay gives the most value to the campaigns that reached the person most recently. Here's how Feathr calculates it:
- Find each campaign's last touch. For each campaign the person interacted with, Feathr uses only the most recent interaction before the conversion. Earlier interactions with the same campaign are ignored.
- Find the oldest touch. Feathr finds whichever of those last touches happened first. This is the starting point.
- Count the days. For each campaign, Feathr counts how many days after the starting point its last touch happened.
- Split the value. Each campaign's share of the value is its number of days divided by the total days for all campaigns. The shares always add up to 100%.
The campaign with the oldest touch is always 0 days after the starting point, so it gets no value and doesn't show the conversion. The date of the conversion itself doesn't change the split. Only the timing of each campaign's last touch matters.
Let's add some timing to the example. Campaign A's last touch was on day 0, Campaign B's on day 5, and Campaign C's on day 9. The total is 0 + 5 + 9 = 14 days.
| Campaign | Last touch | Days after oldest touch | Share of the value | Conversion | Conversion value |
|---|---|---|---|---|---|
| Campaign A | Day 0 | 0 | 0% | — | — |
| Campaign B | Day 5 | 5 | 35.7% (5 ÷ 14) | 1 | $35.71 |
| Campaign C | Day 9 | 9 | 64.3% (9 ÷ 14) | 1 | $64.29 |
Example with five campaigns
The same rules apply to longer paths. Here, a person interacted with five campaigns before making a $100 donation. The total is 0 + 3 + 7 + 10 + 12 = 32 days.
| Campaign | Last touch | Days after oldest touch | Share of the value | Conversion | Conversion value |
|---|---|---|---|---|---|
| Campaign A | Day 0 | 0 | 0% | — | — |
| Campaign B | Day 3 | 3 | 9.4% (3 ÷ 32) | 1 | $9.38 |
| Campaign C | Day 7 | 7 | 21.9% (7 ÷ 32) | 1 | $21.88 |
| Campaign D | Day 10 | 10 | 31.3% (10 ÷ 32) | 1 | $31.25 |
| Campaign E | Day 12 | 12 | 37.5% (12 ÷ 32) | 1 | $37.50 |
Campaign A doesn't show the conversion at all. Campaigns B through E each count it as 1 conversion and get their share of the value.
The exact split depends on when each campaign last reached the person, so it can be different for every person who converts.
Best for: Short decision windows, like a ticket sale or a giving day, where recent campaigns have the most influence.
Keep in mind: The campaign that reached the person first never gets credit under Time Decay. If you want to see which campaigns introduce new people, use First Touch or Positional instead.
Full Attribution
Every campaign in the path shows the conversion and the full value.
In the example, Campaigns A, B, and C each show one conversion and each is credited with the full $100.
| Campaign A | Campaign B | Campaign C | |
|---|---|---|---|
| Conversion | 1 | 1 | 1 |
| Conversion value | $100 | $100 | $100 |
Best for: Very simple use cases, like seeing every campaign that played any part in a conversion.
Keep in mind: Full Attribution credits the same value to every campaign. If you add up your campaigns, one $100 registration looks like $300 in revenue. This makes your ROI look much higher than it really is, so we don't recommend it for ROI or budget reporting.
Compare the models
Here's how each model credits the $100 example registration:
| Model | Campaign A (first) | Campaign B (middle) | Campaign C (last) | Campaigns showing the conversion | Total value across campaigns |
|---|---|---|---|---|---|
| Linear | $33.33 | $33.33 | $33.33 | 3 | $100 |
| First Touch | $100 | — | — | 1 | $100 |
| Last Touch | — | — | $100 | 1 | $100 |
| Positional | $40 | $20 | $40 | 3 | $100 |
| Time Decay* | — | $35.71 | $64.29 | 2 | $100 |
| Full Attribution | $100 | $100 | $100 | 3 | $300 |
*Time Decay values use the timing from the Time Decay example.
A few things stand out:
- Only First Touch and Last Touch show the conversion on a single campaign. With every other model, the same conversion appears on more than one campaign.
- A campaign that gets 0% of the value doesn't show the conversion. This is why Campaign A drops out under Time Decay.
- Every model except Full Attribution adds up to the real conversion value. Full Attribution gives each campaign the whole $100.
Campaign reports vs. Flight and Project reports
Because a single conversion can show on several campaigns, the conversions in your campaign reports and in the Marketing table can add up to more than the number of people who actually converted.
In the example, using Full Attribution:
- Campaign reports: Campaigns A, B, and C each show 1 conversion worth $100.
- Marketing table: Adds the campaigns together, so the one person appears as 3 conversions worth $300.
- Flight or Project report: Shows the person once, as 1 conversion worth $100.
Flight and Project reports remove duplicates. If one person converts after interacting with several campaigns, they count once, with their actual conversion value. This is true no matter which attribution model you choose. Use your Flight and Project reports when you need the true total for an initiative, and use campaign reports to compare how each campaign contributed.
Flight and Project reports have a few other differences, too. Flight reports count conversions against the Flight's goal instead of each campaign's goal, and Project reports combine Flight and standalone campaign results. Read Understanding Conversion Reporting: Campaigns, Flights, and Projects.
Note: The Page totals row in the Marketing table adds up each campaign's numbers and doesn't remove duplicates. Your Project report total will often be lower.
Which model should I use?
There's no single right answer. Choose the model that matches the question you're trying to answer:
| If you want to know... | Try |
|---|---|
| How all your campaigns worked together | Linear |
| Which campaigns bring in new people | First Touch |
| Which campaigns convert people | Last Touch |
| Which campaigns start and finish the journey | Positional |
| Which campaigns mattered most right before a deadline | Time Decay |
| Every campaign that played any part | Full Attribution |
Tip: Try viewing the same report with a few different models. If a campaign shows conversions under First Touch but not under Last Touch, it's good at finding new people but may need a retargeting campaign to follow it.
How to change the attribution model
You can change the attribution model at any time, including after a campaign is published. Changing the model doesn't change your data. It only changes how conversions and value are credited in your reports.
For a single report
Reports are available once a campaign is published. This works for Campaign, Flight, and Project reports.
Step 1: Open the report.
Step 2: Click the ellipsis ( ... ) in the top right of the report.
Step 3: Select Customize.
Step 4: Under Configure report settings, choose an Attribution model.
Step 5: Click Ok.
[Screenshot: Customize Report window with the Attribution model menu open]
The report's Conversion Analytics section shows which attribution model is being used.
For the Marketing table
In Marketing, click Customize columns and choose an Attribution model. This changes how conversions are credited in the table.
For your whole account
Your account has a default attribution model. New reports use this model unless you choose a different one in Customize. Your account default also affects how often you see Recommended for you suggestions in your campaign reports:
- Full Attribution gives every campaign the full conversion value, so campaigns look like they're performing better and you'll see recommendations more often.
- Models that split the value, like Linear, lead to fewer, more conservative recommendations.
When you change your account default, Feathr updates your recommendations automatically. Learn more about Campaign Recommendations.
Common Questions
Does changing the attribution model change my data?
No. Feathr keeps every interaction on the person's path to conversion. The model only changes how conversions and value are credited in your reports, so you can switch between models as often as you like.
Why do my campaigns show more conversions than my Project report?
With most models, one conversion shows on every campaign that gets credit for it. Project and Flight reports remove these duplicates, so they show the true number of conversions. See how each report counts conversions.
Why is my campaign's conversion value lower than the donation or purchase amount?
The conversion value was split with other campaigns. For example, under Linear, a $100 donation shared by four campaigns credits each one with $25.
Why does my campaign show a conversion worth $0?
The conversion itself had no value. For example, the goal may not have a value assigned to it. A $0 conversion still counts as a conversion on each campaign that gets credit for it.
Why doesn't my campaign show a conversion?
The campaign may not be in the person's path to conversion, or the model may give it 0% of the credit. For example, under Last Touch, only the final campaign shows the conversion. Under Time Decay, the campaign that reached the person first doesn't show it. Also check that the person completed the campaign's goal after seeing or clicking its ads.
Do attribution models work with Multi-Convert?
Yes. Multi-Convert decides how many conversions a person can have. The attribution model decides which campaigns show each conversion and how its value is divided. Learn more about Multi-Convert.
Why do my numbers look different across reports?
Check that you're comparing reports with the same attribution model and date range. Also keep in mind that Flight and Project stats stop updating 30 days after a campaign ends. Learn more about conversion reporting.