During a campaign, marketers and clients alike are driven to associate every conversion and its resulting revenue with an action. For instance, was a product’s spike in online orders last night the result of steady TV advertising or the result of a recent direct mailing?

Specifically, marketers want to quantify how much online activity was generated by various offline advertising campaigns. To analyze, we use a process called web attribution by which digital sales revenue is linked to direct marketing initiatives; information designed to aid in determining future spend. There are multiple approaches to attribution, all varying in complexity and accuracy. Regardless of yours, it is a necessary practice. Without utilizing proper methods of attribution, advertisers do not know the true efficiency—the Media Efficiency Ratio—of their offline media.

I think of media relationships as PUSH and PULL. TV and offline media are considered push media. This means you are advertising a message to PUSH someone to call, search online and buy. Online marketing (search marketing) is pull media. Users are searching for your product on Google and other search engines, and you want to PULL them in. The distinction is important because you cannot rely on one form of media alone. You need TV and offline advertisements to PUSH the public to search for your product online, and you need an online presence to PULL them in. Then you need web attribution to tie the conversions together.

Connecting the dots with print media—direct mailings and print advertisements—is generally easier and lends itself to stronger web attribution because printed ads tend to stay in front of your consumer, making it easier for them to go to the unique vanity URL in your ad. Same goes for broadcast TV—the localized nature of the broadcasts makes it easier to analyze changes in digital response activity within a specific DMA.

With national TV media however, you’ll find that marketers often attribute online purchases back to the closest national airing/drop/etc. preceding it, or perhaps connect the digital traffic to events that occurred that day. Attribution by media proximity alone isn’t enough, especially if you have a robust TV campaign across multiple channels. In my experience, the most reliable attribution methodology is a process known as station weighting. This process uses the reach (viewership) of each TV airing, along with its airtime, to determine how much of the digital response was generated by each airing.

Here’s an example of station weighting: Using fictional audience data, let’s hypothetically say a marketer runs commercials on three TV stations within minutes of one another—one on ESPN (79 million viewers/impressions), one on H2 (16 million viewers/impressions), and one on Crime & Investigation (reaching 5 million). The volume of digital activity the marketer receives within minutes of the three commercials is attributed to each airing using station weighting probability. Assume the marketer gets 100 web sales within minutes of all three airings. These three airings account for 100 million viewers/impressions (a nice round number). By weighting the three airings, ESPN is likely responsible for 79 of the 100 orders because ESPN accounted for 79% of the 100 million impressions. Sixteen orders were likely driven by H2 (16% of the 100mm impressions), and the remaining five orders are associated with the Crime airing.

If the Crime airing ran as the last airings of the three, proximity attribution would attribute all 100 digital orders to the Crime airing. Station weighting helps gain far greater accuracy than proximity attribution.

It’s important for clients to feel secure in understanding what drives their online and offline media, and as a product owner or potential client it is important to seek out experienced firms that have both the systems and people in place to provide you with the best integrated media analytics. Here are a few questions to ask as you’re looking for a team that understands web attribution. These questions are designed to give you confidence in the firm’s practices, and will in turn lead you to make the most informed decision for you and your product.

What is their attribution methodology?
What software/database runs this attribution?
Did the agency build software? Or did they license it?
If the agency built it, is the software’s architect on staff?
How flexible is this attribution program? Can they enhance the software to execute functions not included in the current version? If so, what is a realistic timeline?
Is there a dedicated team that manages the software?
What is this team’s organizational structure—e.g., if the agency brings in another client, will they be able to scale to serve you both?
Can they present case studies for how they’ve used the attribution for other clients?
Whatever your approach, web attribution is an inexact science and will never be 100% accurate, but this method has proven to garner the most precision I’ve seen. Additional attribution factors include discounting how much of the digital activity belongs to the TV airings based on its distance from the airings, but the core principal of any methodology should be station weighting.