Unified Attribution Model
Unified Attribution Model: combining aggregated signals and customer journeys
Marketing attribution is one of the most difficult problems for companies investing across multiple channels.
Marketers need to understand which channels are contributing to conversions, revenue, and customer acquisition. But the available data is rarely complete. Some channels are visible in customer journeys. Some are available only as aggregated time series. Some generate impressions, others generate clicks, searches, visits, or platform-specific signals. Offline, marketplace, and privacy-limited channels may not appear in user-level paths at all.
The Unified Attribution Model (UAM) was designed for this fragmented measurement environment.
At a high level, UAM combines two complementary sources of evidence:
- Aggregated channel signals, which describe how marketing activity changes over time.
- Customer journeys, which describe how users move across channels before converting.
The goal is to produce an attribution view that is broader than pure journey-based attribution and more journey-aware than a model based only on aggregated data.
Why attribution needs a unified approach
Traditional attribution methods often rely on one primary data source.
Aggregated-data attribution
Aggregated-data models work with time series such as:
- conversions by day or week;
- revenue by day or week;
- impressions by channel;
- clicks by channel;
- spend by channel;
- searches, visits, email sends, marketplace activity, or other channel-level indicators.
These models are useful because aggregated data is often easier to collect, more privacy-friendly, and available for channels that do not appear in user-level journeys.
They can include channels that are difficult to track at the user level, such as offline media, marketplaces, organic demand signals, or platform-reported activity.
However, aggregated data alone may not fully capture how channels interact inside real customer paths.
Customer-journey attribution
Customer-journey models work with sequences such as:
facebook_ads > google_ads > direct > conversion
or event-level records showing which channels a user touched before converting.
These models are valuable because they reveal the structure of the decision path. They help identify whether a channel often introduces users, assists other channels, appears close to conversion, or acts as a bridge in the journey.
However, customer journeys are often incomplete. They may be available only for some channels, some devices, some browsers, or some users. Privacy restrictions, cookie limitations, and platform silos can make journey data partial or biased.
The measurement gap
In many real businesses, the available data looks like this:
| Data source | Typical availability |
|---|---|
| Aggregated channel data | Available for most or all channels |
| Customer journeys | Available only for some channels |
| Offline media signals | Aggregated only |
| Marketplace activity | Aggregated only |
| First-party journeys | Available but partial |
| Third-party journeys | Increasingly limited |
This is exactly the context where UAM becomes useful.
UAM does not force a company to choose between aggregated attribution and journey attribution. It is designed to combine both when both are available.
What UAM does
The Unified Attribution Model works in two conceptual steps.
Step 1 — Estimate a baseline from aggregated signals
The first step uses aggregated channel data to estimate a baseline attribution across all available channels.
This layer looks at how channel signals relate to the target over time. The target may be conversions, revenue, leads, subscriptions, or another business outcome.
The aggregated layer can use different types of channel signals, such as:
- impressions;
- clicks;
- spend;
- searches;
- visits;
- email activity;
- platform-specific engagement metrics;
- business-specific indicators.
A key advantage is that multiple signals can describe the same channel. For example, a paid search channel may be represented by impressions, clicks, and cost, while a direct channel may be represented by branded searches or direct visits.
This step produces a first attribution baseline that includes all channels represented in the aggregated data, including channels that are not visible in customer journeys.
Two possible views of aggregated contribution
The aggregated baseline can be estimated from more than one perspective.
One view is an Automatic MMM perspective. It is inspired by media-mix modeling, but designed to work automatically with several types of channel signals. It focuses on the predictive relationship between channel activity and the target over time, while allowing different signals to describe the same final channel.
Another view is a Reward Model perspective. Instead of focusing primarily on a media-mix relationship, it looks at how channel movements behave in relation to movements in the target. A channel receives more support when its signal is repeatedly informative with respect to the business outcome.
These two views give UAM flexibility. The Automatic MMM view is useful when the analyst wants an aggregated-data baseline driven by predictive signal strength. The Reward Model view is useful when the analyst wants a more movement-oriented interpretation of how channels behave over time.
In both cases, the purpose is the same: create a robust attribution baseline from aggregated data before adding the customer-journey layer.
Step 2 — Use customer journeys to refine the baseline
When customer journeys are available, UAM adds a second layer.
This layer uses a Markov model to evaluate the role of channels inside observed paths to conversion.
A Markov model is useful because it looks at transitions between channels. It can capture whether a channel often appears in successful journeys, whether it helps move users toward conversion, or whether it plays a structural role in the path.
This journey-based evidence is then used to refine the aggregated attribution baseline for the channels that appear in customer journeys.
The result is a unified view:
- channels with aggregated data and journey data benefit from both sources;
- channels with aggregated data only can still receive attribution;
- the final result remains at the channel level.
UAM is therefore not a replacement for aggregated-data modeling and not a replacement for journey modeling. It is a way to connect the two.
Why this combination is useful
Aggregated models and journey models answer different questions.
An aggregated-data model asks:
Which channels explain movements in conversions or revenue over time?
A journey model asks:
Which channels appear to play an important role inside observed customer paths?
UAM combines these two signals so that attribution can reflect both:
- the macro-level relationship between channel activity and business outcomes;
- the micro-level structure of observed customer journeys.
This is especially useful when only some channels appear in customer journeys.
For example:
| Channel | Aggregated data | Customer journeys |
|---|---|---|
| google_ads | Yes | Yes |
| facebook_ads | Yes | Yes |
| Yes | Yes | |
| marketplace | Yes | No |
| offline_media | Yes | No |
| direct | Yes | Partial |
A pure journey-based model would ignore or underrepresent channels that are not visible in paths. A pure aggregated-data model would ignore useful information from observed journeys.
UAM is designed to use both.
What happens when journeys are missing?
If no customer journeys are available, UAM can still produce an attribution result from aggregated data.
This makes the methodology practical for companies that want to start with the data they already have. A business can begin with aggregated channel signals and later incorporate journeys when tracking becomes available or improves.
This creates a natural adoption path:
- Start with aggregated data only.
- Add customer journeys where available.
- Compare the baseline attribution with the journey-adjusted attribution.
- Inspect which channels are affected by the journey layer.
- Use the unified result to support budget allocation and measurement discussions.
The model can therefore evolve with the company’s data maturity.
What kind of data is needed?
UAM works with two main types of data.
1. Aggregated channel data
This dataset contains one row per time period.
Example:
| timestamp | conversions | google_clicks | google_impressions | facebook_impressions | email_clicks | direct_searches |
|---|---|---|---|---|---|---|
| 2024-01-01 | 120 | 340 | 12000 | 18000 | 210 | 900 |
| 2024-01-02 | 135 | 360 | 12600 | 17500 | 230 | 950 |
| 2024-01-03 | 128 | 330 | 11900 | 18200 | 220 | 910 |
The important idea is that every signal is mapped to a final channel and to a type of measure.
For example:
| Signal | Channel | Measure type |
|---|---|---|
| google_clicks | google_ads | clicks |
| google_impressions | google_ads | impressions |
| facebook_impressions | facebook_ads | impressions |
| email_clicks | clicks | |
| direct_searches | direct | direct searches |
This structure makes it possible to represent multiple signals for the same channel while still returning channel-level attribution.
2. Customer journey data
When available, journey data describes paths to conversion.
Example:
| id_path | timestamp | channel |
|---|---|---|
| 1 | 2024-01-01 10:12:00 | facebook_ads |
| 1 | 2024-01-01 12:31:00 | google_ads |
| 1 | 2024-01-01 12:45:00 | conversion |
| 2 | 2024-01-02 09:22:00 | |
| 2 | 2024-01-02 11:10:00 | direct |
| 2 | 2024-01-02 11:18:00 | conversion |
These paths allow the journey layer to understand how channels interact before conversion.
UAM can then refine the attribution of channels that appear in paths while preserving attribution for channels that exist only in aggregated form.
A high-level example
Imagine a company has the following channels:
google_adsfacebook_adsemaildirectmarketplaceoffline_media
Aggregated data is available for all channels. Customer journeys are available only for google_ads, facebook_ads, email, and direct.
UAM proceeds conceptually as follows.
First, it estimates the aggregated baseline
The model analyzes the relationship between channel signals and conversions over time.
This produces an attribution baseline across all channels, including:
- marketplace;
- offline media;
- channels not visible in customer journeys.
Then, it reads the journey structure
The Markov layer evaluates the role of channels visible inside customer journeys.
This adds information such as:
- which channels are frequently part of conversion paths;
- which channels tend to assist other channels;
- which channels appear closer to conversion;
- which channels are structurally important in observed paths.
Finally, it combines the two views
Channels with both aggregated evidence and journey evidence are informed by both sources. Channels without journey data can still receive attribution from the aggregated baseline.
The result is a more complete view than either source alone.
Why use UAM?
UAM is useful when you want attribution that is practical, flexible, and aligned with modern data constraints.
1. It works with fragmented data
Most companies do not have a perfect attribution dataset. UAM is designed for situations where different channels are measured in different ways.
2. It can include channels outside customer journeys
Channels that do not appear in user-level paths can still be represented through aggregated signals.
3. It benefits from journeys when they exist
When journey data is available, UAM uses it to improve the attribution view for the channels that appear in paths.
4. It supports mixed signal types
A channel can be represented by impressions, clicks, spend, searches, or other signals, depending on what is available and meaningful.
5. It supports staged measurement maturity
A company can start with aggregated data and later add journey data without changing the overall attribution framework.
UAM versus traditional MMM and MTA
| Approach | Main data source | Strength | Limitation |
|---|---|---|---|
| Traditional MMM | Aggregated time series | Works without user-level data | Often requires careful specification and longer histories |
| MTA | Customer journeys | Captures path structure | Requires reliable journey tracking |
| UAM | Aggregated signals + optional journeys | Combines both views | Requires well-structured channel mappings |
UAM does not force companies to choose between aggregated attribution and journey attribution. It allows both to contribute when both are available.
When should you consider UAM?
You should consider UAM if:
- you have aggregated channel metrics over time;
- you have customer journeys for only some channels;
- you want to compare aggregated attribution with journey-adjusted attribution;
- you want to include channels that are not visible in user paths;
- you want an attribution method that is usable under modern privacy constraints;
- you need a bridge between media-mix thinking and customer-journey attribution.
UAM is especially useful for companies that are moving from purely user-level tracking toward a more robust, mixed measurement strategy.
Conclusion
The Unified Attribution Model brings together two complementary sources of marketing evidence:
- aggregated channel signals;
- customer journey paths.
The aggregated layer estimates a baseline view of channel contribution from time-series data. The journey layer adds information about how channels behave inside observed paths.
Together, they provide a unified attribution view that is more flexible than pure aggregated modeling and more complete than pure journey-based attribution when tracking is partial.
If your attribution data is fragmented across aggregated reports and partial customer journeys, UAM is designed for exactly that problem.