With a standard attribution model, a rule decides in advance which touchpoint of a customer journey gets the credit for an order. AI Attribution replaces that rule with a model that has learned from real customer journeys which touchpoints actually lead to purchases.
In this article
Why AI Attribution
Standard models such as Last Click or U-Shape are rule-based. A human decided which position in the journey is important, and that rule is applied to every order in the same way. This has two limits:
The rule reflects an assumption, not the behavior of your customers. Last Click ignores the channels that create demand, First Click ignores the channels that close the sale.
The rule is the same for every product, price point and customer, although journeys differ a lot.
AI Attribution removes the human assumption. No one defines in advance which touchpoint matters.
How the model weighs touchpoints
The model was trained on the customer journeys of all Tracify customers. It uses the interactions of visitors, such as page views, product views, add-to-cart events, session length and the time between sessions, to estimate how likely a visitor is to purchase.
For every touchpoint in a journey, the model estimates this purchase likelihood again. The change from one touchpoint to the next is the contribution of that touchpoint:
A visitor comes to your shop for the first time through a Google ad. The model estimates a purchase likelihood of 10 %.
The visitor returns through a Meta ad, looks at several products and adds one to the cart. The likelihood rises to 60 %.
The visitor comes back through an email and purchases. The likelihood before the purchase is 90 %.
The Google ad contributed 10 points, the Meta ad 50 points and the email 30 points. The order is split in this ratio: Google 11 %, Meta 56 %, Email 33 %. Every order distributes exactly 100 % of its credit, so AI Attribution never counts an order twice.
ℹ️ Note: All values are based on measured events of your visitors. The model does not add estimated or modeled orders, and it does not fill gaps for visitors who could not be tracked.
What is different from the standard models
| Standard models | AI Attribution |
Who decides the weight | A fixed rule (first, last, equal, U-Shape) | The model, based on the behavior in the journey |
Same journey, same result | Yes, the rule is applied identically | Yes, but two journeys with the same channels can get different weights |
Reflects demand creation and closing | Only one of both, depending on the rule | Both, according to their measured contribution |
Totals of the report | Identical across all models | Identical across all models |
The totals of a report do not change between the models. AI Attribution redistributes the credit between channels, campaigns and ads. Overview of all models: What are the Attribution Models?
How you use AI Attribution
AI Attribution is the model of the two default presets in your account:
Preset | Touchpoints | Window | Report time | Use it for |
AI - All Touchpoints | All channels, including Organic and Direct | Unlimited | Conversion Date | The overall picture of your marketing, including organic effects |
AI - Paid Media | Paid channels only | Unlimited | Conversion Date | Optimizing campaigns and budgets |
You select the preset in the dropdown at the top of every dashboard. The browser extension uses the same presets in your ad platforms.
Like every model, AI Attribution works with an attribution window and a touchpoint selection. In AI - Paid Media, the credit of non-paid touchpoints such as Organic or Direct is redistributed to the paid touchpoints of the same journey. A journey without any paid touchpoint does not appear in this preset. Details: How do included touchpoints influence your attribution data?
💡 Tip: You can combine AI Attribution with a shorter window or with selected channels in a custom preset, for example AI Attribution with a 14-day window and Meta only. Contact our support via chat or email and we set it up: Understanding Presets in Tracify
Common questions
Was the model trained on my data only?
No. The model learned from the journeys of all Tracify customers, which makes it robust also for smaller accounts. Your reports are calculated only with your own visitors and orders.
Why does a channel get fewer orders in AI - Paid Media than in AI - All Touchpoints?
AI - Paid Media only contains journeys with at least one paid touchpoint. Journeys that were purely organic or direct are not part of this preset, so the total number of orders is lower.
Can I still use Last Click or U-Shape?
Yes. Every standard model is available as a preset. If your account does not have the preset you need, contact our support.
Why are the numbers different from my ad platform?
Your ad platform counts every order it touched fully for itself and adds modeled conversions. Tracify counts every order once and distributes it across the whole journey: Why can't you track every user anymore?