What is unique for a B2B company when you are building a Marketing Attribution

Content: What is attribution? What is different for B2B? How could it be helpful? Living in a cookie-less world. Identity Resolution.

In the next medium post, I described in detail how to build a b2b attribution data model

Timeline of a particular prospect’s traffic sources and goals. And attribution models contribute weights values

Marketing attribution — the method used to evaluate the value or contribution of different marketing channels toward achieving specific business outcomes, like generating sales or leads. There are:

1. First-Touch Attribution: Credits the customer's first interaction with the brand for the conversion value.
2. Last-Touch: Credits the last interaction before conversion, assuming this was the decisive factor.
3. Linear: Distributes the credit equally across all customer interactions with the brand.
4. Time Decay allocates more credit to interactions closer to the conversion, assuming they had a more significant impact.
5. U-shaped: This method gives credit to the first and last interactions. (sometimes touches in the middle also have some low coefficients)

“first-touch” and “last-touch” are metrics with values of 0 or 1. The picture above shows the most popular attribution metrics calculated for the events. The linear model is just equal to the weight for all events.

What is different for B2B attribution?

• It takes a long time, from the first touch to closing a deal. This makes it much more complicated to understand the causation of spending and efforts to achieve goals. More details below.
• There is a sales process after the marketing job is done. We don’t want a “brand organic search” for the last touch when the client checks the company address while signing the contract. We need some previous goals. I defined it as “marketing goal”
• There are many different goals. It could be dozens. And for every “prospect did something cool” goal, marketers want to check attribution weights.

Insights from Marketing Attribution Metrics. Examples

When I started working with the attribution data output, I felt considerable confusion. Later, I realized that the main reason was two-dimensional time: the traffic source event date and the goal date.

Both dates are essential because the traffic source event involves spending resources, and the goal date involves getting value.

In the chart below, I tried to show how the goal date could be related to the “traffic source event date.” I used the “linear model” for a historical lookback because it shows more nuances.

2024Q1 Close Wons. By attribution model and by traffic source event quarter.

Parametrised things here:

• Sometimes, choose “first won” or sometimes “total wins” for the customer. It could make a difference.
• We can use Revenue instead of Wons. (IMHO, summing revenue could increase noise. The maybe best option is to use wons count and median revenue)

I checked many benchmarks of marketing attribution dashboards and never saw this two-dimensional time comparison. Maybe it's because it’s challenging to visualize. I don’t know. Or perhaps it’s somewhere demonstrated. Please share with me an example.

Return Of Investments. (ROI) Matching Spends Metrics with Attributed Goals

ROI is precisely what the average CEO | CMO | CRO| VP of Marketing expects from marketing attribution.

Again. There is a challenge with dates related to a situation with a significant delay between marketing touches and close won.

Again, I spent a few weeks on data-qa before realizing its data was correct. However, the spend date and revenue date are different.

For business users, it’s the same. It’s confusing that we couldn’t have the same revenue and traffic source event dates. Let me show it as an example. Let’s take a look at how revenue by quarter for paid ads might look like

hypothetical revenue for paid ads, by close won date

We couldn't match the spending above the chart bc spending for ads was on different date ranges.

To compare spend and revenue, we need to switch to “marketing traffic source event date”. But this chart could also lead to confusion for another reason:

Revenue compared to spending by traffic source quarter date

From a finance perspective, it’s weird to look at data with so many variations for spend/revenue metric (someone might call it ROI, but we wouldn’t).

So, what I see in practice is some people might use “date range revenue” instead of “same date range ad spend”. It could work—in many cases, it works pretty well. Example when this approach might lead to wrong decisions:

• CTR, CPC, and conversion from visit to web form didn’t change a lot
• conversion from “Form Submit” to Won dropped

Living in a cookie-less world

some horror news related to the topic

Actually, all this bad news is related only to third parties. This is why leading players already moved to first-party tracking. And problem solved.

First-party web analytics tools:

• GA4 is a first-party, but some traffic may need to be checked. The loss is 0% in some cases and 30% in others. In details. GA4 doesn’t use cookies to store user_id. It saves user_id in server data.
• Mixpanel. Also first-party. But it’s not free. I Love Mixpanel and think it is worth buying if you have a budget.
• Snowplow has an open-source version you can just set up on the server side (first party) and use
• Adobe Analytics ( Adobe Experience Cloud ) says it has a first-party option. However, first-party tracking wasn't set up correctly in all three cases where I checked customers with Adobe Analytics. And people who own Adobe Analytics accounts don’t know how to set it up, so I wouldn't recommend using it.

Identity Resolution

GA4 user_id is a single identity for multiple user devices.

However, using Mixpanel or Snowplow, which doesn’t have this feature, is not a problem. For B2B, this “identity thing” is not worth the time to think about too much.

Next post. Bout how to build attribution data model.

P.S: Like it? Please press like to my github account. It motivates me a lot.