Attribution is a pain point for companies of all shapes and sizes. There are so many options; so many decisions to be made, and there are no clear rules or parameters to guide you.

There are no rules in the sense that there aren’t clear outlines for how to do attribution “correctly” because attribution can be used in different forms to solve different problems.

In this post, we’ll talk about some best practices and how to make your attribution model work for you.

Why does attribution have a bad rap?

Before we get into the meat of this post, we have to address the elephant in the room. Attribution has a bad rap. I’ve heard people go as far as to refer to attribution as “attri-bullsh**”. Others have called it foolish and many believe that it isn’t worth the time. Why?

It’s fair to say that attribution can get messy if:

  • You are using different data sources for all of your different channels, which would lead to some double counting, at best, and a huge disconnect between reporting and actuals, at worst. (It’s often the latter.)
  •  You aren’t accounting for all channels.
  • The lookback window is off. (You’re tying back to touches within 30 days but your sales cycle is 180 days.)
  • Cross-device tracking isn’t in place or accounted for.

Also, there are several different possible models and it can admittedly be frustrating to navigate the different models – and that’s not even considering the difficulty of getting buy-in from all stakeholders to adopt a strategy.

For these reasons, attribution deserves the widespread skepticism and frustration. However, there are ways to deal with these issues if you dedicate yourself to improving your attribution models.

Why does attribution matter?

Consider this a peace offering for whatever despair I may have caused you as you read that last section: I still believe in attribution.

Here’s why: If you are tracking results at *ANY* level, then you are using an attribution model, whether you believe in attribution or not. Therefore, if you’re going to use attribution (and you are), then it should be as accurate as possible.

The goal of attribution is to give better insight into what’s working, what isn’t working and how it all works together. Those are pretty important insights that should serve to inform future decisions. While I’ll be the first to admit that there isn’t a company in the world that has nailed attribution with a 100% degree of confidence, I would also be the first to argue that it’s important to put energy toward making it as accurate as you can.

Now let’s refer back to the list in the prior section. None of these are reasons not to implement attribution. If any of these situations apply to you, you will need to put extra effort into attribution. Arguably, in many cases, a model with parameters and definitions would serve to improve these problems.

Shout out to companies with offline sales cycles & companies that rely on their CRM for reporting

This post is especially geared toward companies that rely on their CRM for reporting. There’s an excruciating need, in these cases, to implement an attribution model.

Why? Because there is so much data and usually so many data sources. Often — but not always — these companies also have a large offline presence. These are also usually the companies that have sales development reps (SDRs) performing targeted outreach in addition to their marketing efforts.

They’re tasked with pulling together all of that data into something meaningful and actionable. The most common CRMs, especially Enterprise-level CRMs, don’t do a great job at making the data cohorts available to analyze marketing data and results — that’s just not what they were built to do.

If you’ve worked with or at a company like this, you probably know how difficult it can be to tie revenue back to marketing efforts. Attribution is important for everyone, but, in these cases, it’s incredibly important because it is the only real way to connect the dots.

Options to explore

There are a multitude of ways you can tackle attribution challenges. Let’s talk through some of the most common ones and I’ll lay out the pros and cons of each.

Google Analytics

Google Analytics is not typically my recommended approach for managing large-scale attribution. However, if all of your sales and marketing efforts take place online, then it may be a good option for you and if you can use the free version, that’s a huge bonus. The other benefit of Google Analytics is that we’re all already using it – if you can get away with using it for attribution without major pitfalls, then that would be a huge plus.

However, be aware that you could be missing data, depending on what you want to count as “touches.” You have less control over what is considered a “touch,” — for example, a link click would be considered a “touch,” which may not always be ideal. You can import offline activity but, for example, an email list that was received from a conference would be hard to track back to that event. Rather the credit would likely go to an email or a remarketing campaign. Another flaw is the inability to track data back to personally identifiable information (PII) – let alone a buying team as a B2B marketer would prefer. If you’re looking for closed-loop reporting, Google Analytics is not the best solution.

I love Google Analytics, but I believe the value of Google Analytics lies outside of wholesale attribution.

Using your marketing automation platform for attribution


Your marketing automation platform may have some capabilities for attributing value back to your marketing channels. When compared to Google Analytics, there are some advantages to using marketing automation. For one, the data is tied to PII, where possible. If you have a long sales cycle, marketing automation can do a better job of visualizing the sales cycle and where prospects are in funnel, currently, in comparison to Google Analytics.

There are some major pitfalls for using a marketing automation platform for attribution, though. For one, most don’t have the ability to look at different models and are not flexible in the way that the data is reported out (Google Analytics does a better job at quantity and flexibility of reporting options). I don’t love marketing automation platforms for attribution but, like Google Analytics, if it is a platform that you’re already using anyway, and you don’t have the resources to invest in a better option, then it’s a start.

Building your own attribution platform


If you have a team of data analysts and a good BI tool, it’s not impossible to build your own attribution methodology, without the support of a platform. That said, there are things of which you should be aware before you try to tackle this:

  • It will probably take some serious lifting. Sometimes just the man-hours alone make it worth spending money on a platform.
  • You’ll be reinventing the wheel for the most part – and you’ll *probably* still have quite an investment in tech although that tech could potentially also be used for other things.
  • The data has to be accessible and formatted in a way that allows you to build cohorts – otherwise you’ll need to implement those processes in order to make your data useful.
  • There are a ton of considerations – which I’ll cover in the next section. This applies to all options but is especially important if you’re building your own platform because, depending on how you’re hoping to look at the data, you may have to determine how to engineer it for your purposes whereas other options involve pre-built solutions.
  • The positive side of this is that you can fully customize it (to the extent that your software allows). You’ll just want to make sure that whatever you build is repeatable.

Implementing a turnkey marketing attribution platform

If you have large, complex data sets from a breadth of different sources, this is arguably one of the best solutions. These are platforms that were invented solely with the purpose of solving these problems. A handful of platforms exist currently, and I expect that number to continue to grow. The biggest drawback here is typically price.

The complexity of the platforms, and the complexity of each individual company’s needs, mean that it is really important to understand what you want to achieve before you begin evaluating these platforms. In the next few sections, I’ll cover things you’ll need to begin considering. The platforms will have some similarities and they’ll each have their own advantages. Creating a decision-making matrix can really help in comparing features and benefits across platforms.

Here’s an example: is it most important to you to have the attribution data within your CRM or is it okay if it lives in a separate platform? (Maybe you would even prefer that it live in a separate platform, as long as it is visual and easy to use, whereas often CRM reporting is not.)

Employing a marketing attribution agency

Last but not least, you can hire someone to help you tackle this challenge. These people are expected to be specialized in attribution – meaning they can help guide you through the process. However, you still have some of the same cons of building a proprietary system in that it will likely take quite a bit of time and money.

On the other hand, it won’t be employee time which is likely already spread too thin. Additionally, you may wind up with a better product. This makes the most sense for companies in one of these situations:

  • The use-case is very unique, and isn’t well-covered by any existing platforms.
  • The methodology could be built to be supported by existing solutions and the cost of the upfront build would be a good investment because it would eliminate the need to license an attribution solution in addition to the existing tech stack.
  • The agency can contribute additional analyses or reporting (or will offer strategy and insights) that are outside of your team’s capabilities.

Features to consider when choosing an attribution methodology

In this section, I’ll cover a few of the many considerations you must take into account when determining your attribution solution. In the next section, I’ll share a list of questions that will help define your selection criteria once you’re ready to move forward.

Tying the data to Personally Identifiable Information (PII)
One feature that I would prioritize is the ability to tie your performance data back to channels (that’s a given) but also back to customer data. Tying marketing touches and performance data back to the customer allow you to have better visibility into the journey. It also gives you the ability to segment the data in new ways, with the other data points that you’ve collected.

In addition, storing all of that information in one place creates new opportunities to reach people in the future. Not only are they cookied from your marketing efforts, but you’ve now also collected more information and associated it with an email address and possibly even a phone number.

An account level view
If you work for a B2B company, there’s a good chance that decisions are made by buying teams at your target companies, as opposed to a sole person. In these cases, it’s ideal to have an attribution platform that’s able to string together touches across an account – including multiple people. If your system isn’t able to track across account contacts, and there are multiple touches by multiple people, each will be siloed. Not only is this an inaccurate view of the purchase process, but it also reflects negatively on the channels that don’t have a purchase associated, since the model is unable to tie it back.

Available integrations
Whenever it comes to martech, integration is key. This is especially true in cases where reporting is involved. Later we’ll talk about defining touchpoints for each channel. However, the way those touches are defined and tracked needs to be accessible to the platform or system that will be used for attributing value. In addition, the system also needs to be able to incorporate revenue. Any information that isn’t integrated or accessible in an automated fashion opens you to risk of error.

Data segmentation capabilities
There are a ton of ways to slice and dice data. As you’re preparing to set up your model, think about all of the insights that you hope to glean from it: sales by product? New vs. cross-sell? Make a list of all of the ways that you’d like to be able to segment your data and then ensure that the set up allows for that level of analysis.

Available models
What models do you hope to have access to? How do you want them set up? For example, if you wanted a model to see last-touch, what would you want to count as the last touch? Last-touch before the lead was created or last-touch before the sale was closed?

It’s a good idea to map out the journey and determine how you want to be able to attribute value throughout the cycle. Getting a good look at the full journey typically requires more than one attribution model – more on that later. The point here is that you’ll want to ensure that whichever methodology you choose to proceed with has the ability to view the data from every angle that you want.

Questions to answer when choosing an attribution platform

Following, a list of questions that should help you get started to ensure that you know what you hope to achieve before you get started. Answering these questions will help you decide.

  • Who will be responsible for maintaining the system – both from a technical perspective as well as data governance?
  • Who will be using the system on a regular basis to pull data? What kind of training and/or access will be needed to support that?
  • What do you want to consider as a first touch? The very first touch ever, even if it was years ago? Or a first touch after a lead has been recycled or otherwise disqualified from the last sales cycle that it participated in?
  • What do you want to consider as the last touch? The last touch before the sale? The last touch before the demo? The last touch before the MQL?
  • Can this system track every potential channel that you need to track?
  • Where does the data live now and do the appropriate processes or integrations exist in order to compile it in one place?
  • Can this system track every potential touch? Are you able to define what a “touch” is?
  • Do you need to track broadcast marketing efforts?
  • Is there a cut-off or lookback window that you want to enforce?
  • Can this system look at account-based attribution or only individuals?
  • How do you hope to be able to segment the data in the future? Examples could include:
    • Pipeline and Revenue for New Sales vs. Client Sales
    • Pipeline and Revenue by Business Line or Segment (SMB vs Enterprise)
    • Pipeline and Revenue by Channel Groupings
    • Pipeline and Revenue by Individual Campaigns
  • Can this platform help you to understand funnel conversion and lead velocity? Better yet, can you achieve this by channel?
  • What attribution model(s) do you want to be able to see?
  • Do you intend to create a custom attribution model?
  • Who should have access to all of the data?
  • Can you filter out bloat from junk leads?
  • How do you want date ranges to work? For example, if you choose a date range, do you want to see only leads that were starting the sales cycle no matter when they closed? Do you want to see only leads that closed in that date range no matter when they started the sales cycle? Do you want to see only leads that entered the pipeline in that range – no matter when they entered the sales cycle and no matter when they closed? Do you want to see every sale that was impacted by a marketing touch during that time frame, no matter if it was the first or last touch or anything in between? Note: there are distinct purposes for each of these controls. If you wanted to look at the quarter’s revenue performance the way that the sales team is looking at it, then you’d want to look at sales that closed in that quarter, despite when the campaign occurred. But if you wanted to look at that quarter’s campaign performance, then you’d want to either look at leads that had their first touch in that quarter or all leads that had a marketing touch in that quarter – depending on what questions you were hoping to answer.
  • What questions do you hope to answer about your pipeline, your revenue sources and your campaigns? Are there any other questions that other stakeholders (even from other teams) that could and should be answered with the same data?

Implementing a formalized attribution plan

Determining what to track: Teamwork makes the dreamwork
Typically multiple teams will benefit from an attribution model, so it’s a good idea to involve all stakeholders from the beginning. The only way for an attribution model to succeed is if all stakeholders are on board. Otherwise, reports will lead to arguments and mistrust, which is the worst case scenario. Unfortunately, the investment in marketing automation is almost entirely lost if people don’t believe in the reporting.

The goal should be to have all stakeholders reporting out using the same definitions and parameters, which isn’t an easy task. In order for this to happen, the stakeholders need to meet and decide what “touches” will be included in the attribution model. For example:

  • When attributing email, will you count each email send as a touch? Each email open? Only link clicks? Something different?
  • For outbound calls, will you count each call as a touch? Only those that answered? Only those that scheduled a follow-up right then?

Every single engagement that will be counted as a touch needs to be defined, as does the way it will be measured.

If you don’t include all key stakeholders in these discussions, inevitably one of them will not agree with the way that value is attributed to their team’s efforts and it will result in mistrust of the data. The biggest tip that I can give you is to start these conversations early. If everyone is happy with the way the touches are set up, you’ll have a much easier time introducing the reporting when it is ready.

Which attribution model is the right model?
I am often asked which attribution model is the “right” model. I suggest thinking of attribution models as maps. Each map is different – that doesn’t make one less valid than the other – but the one that you use varies depending on what you are looking for.

Picture yourself in the middle of a mall looking for a particular store – do you pull out an atlas? Heck no! Likewise, you wouldn’t consult the mall directory for vacation hotspots.

For more information about attribution models and when to use each, check out this post.

Much like the collaboration that must occur to decide how to track each touch, it’s important to ensure that everyone pulling reports is following the same methodology for time frames, cohorts, and models. The easiest way to achieve this is to work together to develop definitions and reporting processes.

Changing the conversation about who gets credit
Attribution models require the internal support of all stakeholders. Naturally, the conversation in most discussions either starts with, or ends with “so which channel (team) gets credit,” which can create tension. What the real question should be is “What do we need to do to get closer to our goal?” with everyone focusing on playing their role to achieve the common goal.

Although the goal of attribution is to help discern performance, the dialogue has to be protected. Conversations have to stay productive. Instead of playing the blame game, figure out where channels are working (First touch? Middle touch? Last touch?) and where they aren’t.

Figure out which pieces of content are working, which sources, mediums and campaigns are working and which aren’t. Then use all of that information to find insights on how to make them work better, instead of focusing on the negative. Sometimes tough decisions have to be made, to cut out certain channels and reallocate budget. But those conversations should be productive and constructive – not personal.

If the attribution model and the insights begin to feel personal, teams will start to reject the data. That’s why it is so important that everyone agrees to the model at the outset and that everyone is working together to achieve a common goal, objectively.

TL;DR summary

If you’re interested in setting up closed-loop attribution, you have a few options and each has pros and cons. There are a number of factors that you should consider, depending on what you are trying to achieve. Most importantly, no matter which solution you decide to move forward with, the key to success will be in ensuring that all stakeholders are on board with the parameters for measurement, the way that reports are pulled, and the way attribution is discussed.

Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.

About The Author

Amy has built and implemented multichannel digital strategies for a variety of companies spanning several industry verticals from start-ups and small businesses to Fortune 500 and global organizations. Her expertise includes e-commerce, lead generation, and localized site-to-store strategies. Amy owns Cultivavtive, a performance marketing agency.



Sumber: searchengineland