One of the most powerful dimensions of digital advertising is the ability to track individual users as they traverse the internet. Through the use of suitably tagged digital ads, browser technology, and ad server technology, advertisers can trace individual users across websites over time. These resulting data paths, sometimes referred to as streams, have a wide range of uses: profiling, segmentation, and most notably, measuring ad effectiveness. This latter use is a process known as attribution. Attribution is the process of identifying a set of user actions or events that contribute in some manner to a desired outcome and then assigning a value, usually represented as a dollar amount, to each of these events. Marketing attribution provides a level of understanding of what combination of events in what particular order influence individuals to engage in a desired behavior, such as a purchase or conversion.

A Note on Privacy

In nearly all cases, digital ad tracking refrains from directly collecting any Personally Identifiable Information (PII), because it often is not needed. There are some vendors who attempt to infer certain demographic characteristics of particular streams, but this is often done by inference from the stream itself. For example, if it's known that 90% of visitors to a site are within 25-35 in age (usually verified through visitor surveys), they will guess that all ad streams that display an ad on that site are very likely to be 25-35 in age. Users are generally in control of how well they are tracked online (thanks to privacy laws such as GDPR & CCPA). If you want to see this in action, I recommend checking out your Google's ad profile on yourself here, assuming you have ad personalization on.

Why care about Attribution?

The purpose of marketing attribution is to quantify the influence each advertising impression/action has on a consumer’s decision to engaged in a desired behavior (e.g. register or make a purchase). Visibility into what influences the audience, when and to what extent, allows marketers to optimize media spend to maximize the desired behavior and compare the value of different marketing channels, including paid and organic search, email, affiliate marketing, display ads, social media and more. Understanding the entire conversion path across the whole marketing mix provides a distinct advantage over analyzing data in siloed channels given that a typical consumer will cross multiple channels. Usually, attribution data is used by marketers to adjust or plan future ad campaigns by analyzing which media placements (ads) were the most effective towards a specific goal.

Multiple Approaches to Attribution

There are multiple different ways to approach assigning attribution:

• Single Source Attribution (also Single Touch Attribution) models assign all the credit to one event, such as the last click, the first click, or the last channel to show an ad (post view). Simple or last-click attribution is widely considered as less accurate than alternative forms of attribution as it fails to account for all contributing factors that led to a desired outcome.
• Fractional Attribution includes equal weights, customer credit, and U-curve models. Equal weight models give the same amount of credit to the entire media mix, customer credit uses past experience and guesswork to allocate credit, and the U-curve assigns all the credit to the first and last touch, discounting what happens in the middle of the stream.
• Algorithmic or Probabilistic Attribution uses mathematics and statistics to assign conversion credit across all touch points in the stream. Algorithmic attribution starts at the event level and analyzes both converting and non-converting streams across all channels. Weights are then combined by grouping such as placement, site, or channel as data granularity is decreased, allowing the data to point out the hidden correlations and insights within marketing efforts.

What about Offline Media?

Television ads, radio ads, print media, and billboards are examples of offline media that fundamentally cannot be directly tracked. This poses a challenge to any marketing attribution model. How would you account for the influence of these ads in which you cannot track? In reality, this behavior can be fuzzy match tracked; but it's much more a science than a precise measurement. For example, television ad times are well known down to the second, so marketers can look for any corresponding rise in website visitors right around that period of time. Further, through internet service providers (ISPs), an approximate geographical location of where you are is known (usually no more detail than which city you are in). If say a newspaper has a large print subscriber base in that city, marketers may fuzzy match that you were more likely than not to have seen that print ad. So while it is very difficult, there are many tricks to try to do this fuzzy match for offline media. As a result, it is still possible to assign some level of attribution to even offline media.

So What? How does this help?

At the end of the day, attribution assigns a value, usually in dollars, to each advertising channel. Marketers know exactly how much it cost to do advertising in that particular channel. All put together, this provides a clear picture as to which channels are more profitable where the value is far greater than the cost to market, and which are not. This near real time feedback to allow marketers to quickly optimize their marketing channels on an on-going basis.