Engagement-Based Attribution: A Deep Dive into The Most Accurate Way to Assign Value to ABM

The problem of Marketing Attribution has been challenging marketers ever since there’s been marketing. We want to be able to measure and evaluate the effectiveness of our marketing investments to business results – and get precise measures of ROI. Of course, we’re never going to have the same direct connection to business results that our colleagues on the sales team have. Marketing “influence” is more indirect – and has already required a more sophisticated analytical approach.

This isn’t to say that Marketing should be trying to compete for resources with sales or even using ROI analysis to “prove” marketing works. Don’t fall into those mental traps!

Insight opportunity cost

When you spend time and energy justifying marketing, you waste time and resources. “Proving” marketing has an insight-opportunity cost. Directed elsewhere, that same effort can create insight to improve marketing.

Most companies find profits increase when constrained analytics resources focus on improving key decisions, rather than proving Marketing deserves to exist.

Instead, the best Marketing teams use ROI analytics to “improve” decisions about where to focus resources and about how to best work with sales to drive results.

So, why is attribution so hard for marketers?

Attribution is Hard

74% of B2B marketers can’t measure or report on programs’ business contribution! — ITSMA/Visionedge Marketing

Done right, analytics offer a competitive advantage over three-quarters of the companies out there.

“Did this program deliver results?” In B2B marketing, this question is easier asked than answered for various reasons:

  • Long sales cycles: programs often drive pipeline and revenue after months or years.
  • Multi-channel interactions: customers interact with businesses on multiple channels, online and offline. Connecting channel specific identities is tough.
  • Large buying committees: it’s hard to understand program impact across the increasing number of decision makers in accounts.
  • Extraneous variables: non-programmatic factors affect the ROI of marketing programs, including marketing dynamics, macroeconomic trends, and sales rep effectiveness.
ROI is EASIER when… ROI is HARDER when…
Customer journey is 100% online Customer journey is online and offline
Single marketing channel Multiple channel interactions
Single buyer Buying committees
Purchase takes minutes Purchase takes months or years
<$1K deal size >$10K+ deal size

Basically, the very factors that make ABM necessary also make ROI measurement hard.

This brings us to attribution.

In marketing, attribution is the identification of a set of user actions (“events” or “touchpoints”) that contribute in some manner to a desired outcome, and then the assignment of a value to each of these events.

wiki gio

Let’s deconstruct this definition in the context of B2B marketing.

  1. Set of user actions (“events,” “touchpoints” or ”activities”)
  2. Desired outcome
  3. Assignment of a value

This is the 3 part recipe for attribution excellence. If you’ve got all three of these elements, then you’re on your way to answering some complex marketing questions.

In the rest of this post, we’ll dive deep and explore each element of our definition of attribution in detail.

Deconstructing Attribution, Part 1: “A Set of User Actions a.k.a. Engagio Activities”

Sometimes called “touches” or “events” or interactions, these are all the different customer actions that we know about and typically track.

What type of actions do your customers take in evaluating your solution?

This is where you should be thinking about things like:

  • Website visits
  • Conferences
  • Webinars
  • Emails
  • Advertising
  • Sales Meetings
  • Content downloads
  • Etc.

This is where things start to get complicated and you run into the issues we mentioned earlier: you have large buying committees interacting with your content across many channels and over a long period of time. These “whale accounts” are not easy to navigate.

hunting whales

Critical account interaction data is scattered and siloed across multiple sources:

  • Your CRM. Accounts live alongside everything else in the CRM: opportunities, activities, campaigns, not to mention leads and contacts.
  • Your marketing automation. This data includes people and programs, as well as digital behaviors like email opens and website visits.
  • Your email. Until you integrate valuable information from your corporate email and calendar, your analytics lack critical account-level visibility.
  • Your website. You betcha your prospects are on your website – but do you know what brought them there and what they’re reading?
  • And more.

This is where the Engagio Foundation really becomes a crucial source of all these different engagement activities or actions that are organized around specific prospects or customer accounts:

ABM Foundation

This is the main dataset of Marketing and Sales engagement interactions that we’re going to be using for our attribution models.

Not only is this data in one place, but it’s also scored with Engagement Minutes.

What is an Engagement Minute?

Engagement describes something fundamental about the customer’s connection to your brand:

  • Higher degrees of engagement means a deeper commitment
  • More time
  • More emotion
  • More of a relationship
  • More activity – such as buying and advocating

At Engagio, we use Engagement Minutes to track the time your accounts are spending with you. These minutes should cover when they respond to your marketing programs, but also when people interact socially, use your product, and talk with the sales team.

By combining these interactions at the individual and account level, you get a good proxy for engagement.

You can say things like – target accounts spent 2,219 minutes engaging with marketing activities this quarter, up 122% from 1,932 minutes last quarter. And then you can drill into the data to show the specific sales territories, industry segments, and personas with the biggest growth.

This is certainly a sign that things in the middle of the funnel are progressing in the right direction! With this insight, you can identify Marketing Qualified Accounts (MQA’s). You want to identify which are the most engaged accounts and alert sales about any accounts whose engagement is spiking up or down versus their recent trend.

At Engagio, we think that account-level engagement is a better indicator of potential buying activity than individual lead scores which is why we use Engagement Minutes.

what is engagement

And yes, you do have to solve for L2A, and many other technical details around “how” a database like this is possible. Also, if you have a consolidated database like this you can use it for many other useful things in sales and marketing.

Let’s take a look at this dataset in more detail. What should an individual marketing activity look like?

Anatomy of Marketing Activity

A “marketing activity” what you might think it is: the data should answer Who did What and When, and the more details you can capture the better.

anatomy of a marketing activity

Obviously, different data sources will have different levels of granularity here, so this brings us to the following list of data requirements:

Marketing Activity Data Requirements for Attribution:

  • Historical. The dataset can’t just reflect the state of the data today – it has to be historical so that we can “recreate” how a deal comes together over a period of months or years.
  • Organized and Consistent. Think through your naming conventions, marketing channels, engagement statuses, and all other data that will be aggregated and displayed in dashboards.
  • Multi-Channel. The data has to come from many different channels, not just CRM, not just Marketo – but from all the channels where your customers are.
  • Account Based. (Leads and Contacts) Just because Salesforce doesn’t make it easy to keep track of all the people associated with a specific customer account, doesn’t mean your data should be properly consolidated.

Ok, now we’ve got our killer marketing + sales activity database. Let’s start looking at it against the business outcomes we’re hoping to drive.

Deconstructing Attribution, Part 2: “Desired Outcomes a.k.a. Business Results”

How do you measure results in your business?

Is it revenue? How about profits?

How should B2B marketers measure success?

Most metrics that might seem obvious to a casual business observer may not be an excellent measure for B2B marketers because metrics like revenue and profits are lagging indicators of business results.

We already established that in B2B marketing deals take a long time to develop. By the time a deal is won, and money is the bank, marketers may not have an opportunity to learn lessons about which campaigns were successful vs. which were not.

Leading Indicators Lagging Indicators
  • Predictive of future success
  • Allows time for course-correction
  • Less controversial indicator of the objective’s success
  • Generally more easily captured
  • Based on hypothesis of strategic drivers
  • Link to outcome maybe harder to understand
  • Too late to effect the outcome
Examples in B2B Marketing MQLs, MQAs, Meetings, Demos, # of Opps, Sales Pipeline Revenue, Profits

“So Grant, are you saying we should stop thinking about Revenue Attribution?” you may ask.

No, what I’m saying is that is when looking at revenue attribution, it will be more of a reflective exercise where you’ll stand back and say “hmmm, so that’s what happened.” But it will be a lot less actionable because it already happened. There is nothing else you can do, especially in marketing, to impact that revenue. Books are closed. Literally!

It’s better to use revenue attribution to confirm what you already know or suspected, and to then change your marketing plan for the future if you find something you weren’t expecting.

Imagine launching marketing campaigns and then having to wait for months – or even years – for the results. It’s simply unacceptable to expect marketers to have to wait that long to get any meaningful feedback about which marketing tactics work vs. which don’t.

So, it’s therefore helpful to define “top of the funnel” metrics and middle of the funnel and bottom of the funnel metrics.


It’s very easy here to get lost in the shuffle here and confuse everyone on your team – because you can name dozens of measures – from # of leads, to meetings and MQLs, SALs, etc. Don’t fall into this trap! You need a consistent system to organize your data.

I recommend that you furthermore organize your metrics around the Account Journey:

account journeyNow we’ve got a framework for measuring results. It’s about how the Account is moving forward in the sales and marketing process and follows consistent stages.

You should define what these stages are in your organization (feel free to use our examples), and get alignment from all of the stakeholders around this yardstick for measuring success.

Our job then, for attribution, will be to be able to say things like “to drive accounts to stage 2, these are the sort of activities that make a difference.”

I want to focus on one measure specifically which is a favorite for B2B Marketing attribution – and that’s sales pipeline.

Sales Pipeline is a good measure of success to start with attribution because:

  • It’s usually well-defined within your org already
    • Both sales and marketing teams can align around it
  • It’s sufficiently down the funnel for marketing to be driving “business results”
  • It’s sufficiently up the funnel for marketing to be able to have meaningful influence

If you’re just getting started with all of this, I would recommend focusing on sales pipeline numbers for attribution. Most of the examples below will be for this use-case.

Deconstructing Attribution, Part 3: “Assignment of Value a.k.a. Attribution Modeling”

Ok, so let’s recap: we’re going through the 3 elements that make up “Marketing Attribution” and we’ve now arrived at part 3: “Assignment of value”:

  1. ✅ Set of user actions (“events”, “touchpoints” or ”activities”)
  2. ✅ Desired outcomes (business results)
  3. Assignment of a value (attribution modeling)

Finally! After all the setup, we now are ready to dive deep and talk about attribution more directly.

What does it mean to “assign value”?

We have a set of user actions (spanning a period of time, multiple users, multiple channels) and a business result. The questions now are:

  • How did those actions influence the result?
  • Which of the user actions had the most impact on the result?
  • Which user actions could we drive more of in order to drive more results?

The assignment of values – or attribution weights to each user actions – let’s determine which actions had a more significant impact, and therefore should get more of the attribution “credit” for the result.

Here’s an example. Let’s say we have an account with 1) A Set of User Actions and 2) Desired outcome (i.e. a “Purchase” in this case):

assignment of value

The first order of business here is to collect and organize this data. BUT WE ALREADY DID THAT in part 1! That’s what Engagio’s database is all about:

Not every marketing touch is meaningful enough to be able to say that it had an influence on the outcome. We’ve essentially become too good at tracking interactions, and now need to decide which actions we’ll count and which we’ll throw away.

How should you decide which interactions are not really a touch and which ones should stay?
At Engagio, we think that any touch that counts should be customer driven (i.e., did the customer do something vs. did we do something to the customer?) For example, Email Sent by itself is not a touch, but Email Clicked could be (unless it’s a click to Unsubcribe!)

Think of this as separating signal from noise. Engagio makes this part easy with out of the box configuration to get you started.

So, let’s take all that user activity data and parse out Touch vs Not a touch:


Now we can start to assign “Value” to each of these activities using attribution models:

attribution model

Here’s what I have to say about each one of those models:


Attribution Model Formula Interpretation Pros Cons
First Touch Model Assigns 100% of the value to the first touch. What matters most is the very first touch with the Account – that’s what should get 100% of the credit for the business outcome. Identifies programs that source the best leads and accounts.

Relatively easy to implement.

Creates “investment per lead” insight.

Ignores all touches but the first.


Oversimplifies and inaccurately portrays long B2B sales cycles.


Overweights lead generation.


Underweights marketing nurturing and sales touches.

Last Touch Model Assigns 100% of the value to the last touch. What matters most is the very last touch prior to the business outcome – and it should get 100% of the attribution. Shows touches that move accounts to particular stages – i.e., what was the last touch prior to MQA, or Meeting or Opp or whatever stage you care about. Ignores all touches but the last.


Is unrealistic and inaccurate for complex B2B sales.

Equal Touch Model Assigns equal % of the value to all the touches. All of the touches matter equally – and all contributed to the business outcomes equally. The most egalitarian of the models! Evaluates touches across entire accounts not individuals.


Appropriate for lengthy revenue cycle with many touches.


Avoids assumptions about weighing each step.

All touches don’t actually have the same value.
U-Shape (Position Based) Model Assigns fixed % to the first and last touches, example 40% and then splits the rest evenly. First Touch counts and Last Touch counts a lot – but all the touches in between count too! This is the indecisive model. Gives most credit to lead generation and conversion, the two priorities execs tend to care about most. May overlook channels, programs, and content that encourage the customers mid-funnel (eg nurturing).

Here’s what it looks like if you plot the values in a chart for each model:

traditional attribution models

Based on all the different conversations I have with Engagio customers and prospects, and everyone else that I’ve talked to over the years, I would say the most common attribution models that we use today in B2B marketing is… no attribution model at all!

In other words, we don’t even get to this point because it’s just too hard to do all the 3 steps we’ve outlined here. But of those organizations that do look at multi-touch attribution the most common model is the Equal model. It’s the easiest to execute and easiest to explain.

But we can do better than this.

In part 1 we discussed Engagement Minutes. Surely the “level” of engagement matters here. Surely a single web visit or an email click doesn’t have the same level of influence as an all-day customer workshop. Or how should you compare a sales demo conducted with the key members of the buying committee vs a white-paper download?

The minutes provide us with the language to begin to talk about each marketing touch and help us compare interactions across multiple people and across multiple channels.

So, we should use Engagement Minutes in an attribution model as well:

engagement minute attribution

The more minutes the touch has, the more weight it should have in our model.

Here’s the math for it and comparison with the other models:

engagement minute attribution model

This model that we’re calling Minute Based Attribution Model builds on the previous models but adds a key new element: a powerful new activity scoring mechanism. It gives the model countless customizability based on the type of activity, channel, content, timing, persona, account, etc.

It provides fascinating results that combine information that previously used to be hidden.

engagio dash


You’ll notice that the dashboard above is in Salesforce, which means how you view the data is really up to you. You can customize the charts/reports as needed and drill-in to the level of detail required.

You’ll also notice that the model captures activities like email opens/clicks, web visits (anonymous + known), marketing programs and sales activities like meetings and sales emails.

I like to start all attribution dashboards with a single number that represents what we’re going to be attributing. In this example, it’s $15M. That’s the sales pipeline that we’re going to be attributing in this example – and it comes from a report of Opps in Salesforce. Our job in the dashboard is to help explain the factors that help influence this number (i.e., do attribution).

Right away we take a look at the break-down of engagement between marketing and sales activities and might see that it’s a like a 60/40 on the attribution with 60% marketing and 40% sales.

We also show the average # of touches (Sales and Marketing) prior to the creation of the opportunity. I intentionally made the number be really high at 219.4 touches because that’s what I see in real life. Remember we’re counting web visits and email interactions here! And also notice that we’re on average talking to 9 people at the account. You could do the math for interactions per person and get roughly 20 touches per person per Opportunity, which is pretty realistic, especially stretches across 251.7 days (from the first interaction and the creation of the opportunity).

pipeline attribution screenshot

In the demo dashboard there is a category called “Web – Paid Advertising” that captures web visits (again, both anon and known) that were driven by ads – and captured with UTM parameters. Yes, of course, you can drill into the campaigns and sources – the image just shows the roll-up.

Also, in the dashboard, if you look closely, you’ll see a category for Intent Data, which is an interesting category because it’s not a direct interaction with Marketing or Sales – and instead represents something else entirely. I think it’s essential to include intent data in attribution because we do invest in this data and act in response to it, and technically we’re able to add it to the model. But I’d love to hear your thoughts on how we should categorize it – it’s not really Marketing Engagement nor is it Sales Engagement. What should it be instead?

Key Takeaways

Remember the 3-Step-Attribution formula we covered here:

Attribution = 1 + 2 + 3

  1. Set of user actions (“events”, “touchpoints” or ”activities”)
  2. Desired outcomes (business results)
  3. Assignment of a value (attribution modeling)

And keep these 3 tips in mind when building your attribution models:

  1. If you’re in a complex B2B org (large buying group, takes a long time to buy) you must be able to track account-based attribution.
  2. Engagement Minutes are a super helpful tool to measure engagement across many different marketing and sales activities.
  3. Telling marketing stories supported with data is worth the effort.


Grant Grigorian
Grant Grigorian
Grant Grigorian is the Director of Product Management at Engagio. Grant is on a mission to help marketers make better business decisions based on data. He believes that bringing more science into the art of marketing makes both marketing and sales more effective, and drives business performance.

Leave a Comment

Your email address will not be published. Required fields are marked *