One of the hardest parts of marketing is figuring out where your conversions actually come from and which campaigns should get the credit. Let me end the argument right now: give all credit and money to your content writers.
Just kidding. You should build a marketing attribution model, and then give all the money to your content writers.
Conversion paths grow more complicated with every new type of touchpoint. Gone are the days of simply tracking conversion paths through the telephone number listed in a print ad. Neither your conversion paths nor your customer-brand interaction map boil down to one or two touches anymore. You have to distinguish between direct, organic, social, email, paid, and any outside sources to get a true picture of your customer behavior. Once you see how customers interact with your brand, you can begin to affect that behavior.
ALSO READ: 6 Moz Alternatives for Any SEO Project
How do you know how customers interact with your brand? Data. Every piece of content, social share, phone call, and email has the potential to send data back to your analytics tool, as long as you tag each of those potential touches.
Tagging tells your analytics tool where data comes from, and then you can see the full picture of the customer journey, right? If only it were that easy.
Marketers and data analysts use a number of different marketing attribution models to test their theories on customer behavior. That means no model will work the same for all industries. Warning: I’m not going to give you a model that works right out of the box at the end of this article, because that model does not exist. Here’s what you’ll learn instead:
- The types of attribution models that come standard in Google Analytics (and other, similar tools)
- What works and what is trash about those models
- The best models for beginning your research
- Important things to consider before you get started
A Word on Direct Traffic
Don’t be fooled by sessions labelled as “direct” in your analytics tool. There are a couple of things about the way that direct traffic is calculated that every marketer should know:
- According to Bizible, “Traffic attributed to Direct is typically defined by marketing analytics as any time a visitor manually enters your URL. But in reality, just about every marketing analytics product considers any visitor who doesn’t have a referral source as Direct.” In other words, you may be getting artificially inflated visitor counts.
- SSL matters: Google and other search engines count traffic that moves from an HTTPS secure site to an HTTP site as direct traffic. Customer referral data is stripped when the user moves from a secure to a non-secure site, so you may be losing a lot of really helpful referral data, and without referral data, your analytics tool labels that traffic as direct. See the problem?
Two ways to fix this. You should do both:
- Get an SSL certificate on your site.
- Use UTM tags for all of your campaigns. </rant>
Marketing Attribution Models
Think of your marketing funnel like a group project in school. Everyone puts in different amounts of time building the project, and some folks do much more work than others. Attribution models try to give credit to the team members that do the most work, but most of these models are flawed. You may find some of these useful for edge cases and outliers, but most of them don’t work for real-life attribution. Let’s take a look.
Don’t use this model; it gives all the credit to the last touchpoint before conversion. That means that whoever turns in the report gets all the credit for the work. On the surface, last-touch attribution seems like it makes sense, because the final touch (the last piece of content or ad or whatever) gets all the credit. But this model doesn’t take into account all of the other touches that brought the customer to your brand, and it assumes that the last touch is what made them say, “Yes, I want this.” and click the buy button.
Just like the last-touch model, except it credits everything to the first touch. Still not good, and for the same reasons: it ignores all the other touches along the buyer’s journey.
Last Non-Direct Click Attribution
This is the Google Analytics default, and it skews data just as badly as the first two. This model gives all credit to the last campaign (social, email, paid, whatever) before a direct visit to your site—when the customer types your site into the search bar. Using the group project analogy again, this model is akin to saying the person who printed out the report and gave it to the hander-in gets all the credit, even if that person didn’t do any other work on the project. It’s unfair for everyone.
Avinash Kaushik of Market Motive provides a little bit of insight into how this model trashes some of your Google Analytics reports: “This model is . . . the irritating reason why none of your standard Google Analytics reports match your standard Multi-Channel Funnels reports, even if you look at conversions in the standard MCF Overview or Assisted Conversions reports.” Thanks, Google.
Last AdWords Click
There’s nothing wrong with attributing all of your marketing efforts to Adwords.
Okay, sorry. I can’t keep a straight face. Obviously, this model is just as problematic as any of the others we’ve discussed so far. Why in the world would you ascribe all credit and power to the AdWords algorithms? This model might be valuable if you’re trying to understand which parts of your paid campaigns are working, but then again, you can access this same data through your actual AdWords account.
Everyone in the group gets the same credit, no matter how much work they did. Great for the guy who goes out for nachos instead of meeting with everyone else in the library, but not great for your hard-workers. Also not good for trying to figure out which of your touchpoints bring in conversions and which ones drain your resources.
Time delay actually makes a little bit of sense, unlike all the others we’ve seen so far. Yes, you should give more weight to the clicks that are closer to conversion (this is where our group project analogy falls apart a bit, but if we assume that the people who are most involved in the project are also the ones printing and handing it in, then it goes over smoother). If you focus your efforts on the parts of your funnel that drive conversions, this model will help you clarify which campaigns move customers to buy. But better models exist. Keep reading!
This one automatically gives 40 percent credit to both the first and last interactions, and then evenly distributes the rest of the credit among the middle interactions. This model works better than first and last-click attribution, because it gives a little credit to those campaigns in between, but it’s still not great. Not as good as the time delay, for sure.
Not for beginners, and not for the casual marketer (as if those exist). Tread carefully here, and plan ahead. Just because this is the best model in most cases doesn’t mean that everyone should jump right into it. Nevertheless, when you’re ready, we’ve put together some best practices for building a custom model in the section below.
Building your Custom Marketing Attribution Model
In the interest of avoiding unpleasant surprises, go through this checklist before you build your custom model:
- Tag your campaigns. All of them. If you’re not already using UTM parameters for your campaigns, stop what you’re doing and build those. Remember: Google Analytics’ attribution works with these tags in mind (Google Analytics acquired Urchin in 2005, as in Urchin Tracking Module [UTM], because their tagging parameters and analytics model made sense). You can lean on Google’s Campaign URL Builder until you get familiar with the structure.
- Check your CPA data. Cost per acquisition helps you understand how much you spend on each conversion and will help you decide where you’ll put those funds in the future.
- Count your micro-conversions: How do users interact with your brand? Those big conversions live at the end of the cycle, but count all the other ways users interact with your brand: whitepaper downloads, clicks on blog articles, email sign-ups, social shares, webinars, anything.
- Question the types of micro-conversion behaviors that show the most value for your brand. This takes a little bit of intuition, a little bit of common sense, some customer psychology, and an awareness of your own biases. Talk this out with your team, as the SEM expert and the content writer will have different views and biases.
- Build an accurate time frame for conversions. Do you really know, based on data, how long it takes your customer to convert? Thank goodness Google Analytics has this answer ready-made for you. Navigate in your Google Analytics to Conversions>Multi-Channel Funnels>Time Lag. This shows you the normal rate of interactions with your brand on your site. You’ll probably want to pad this a bit to account for outliers. Also, if you remember from the beginning of the article that tidbit about direct traffic, you may need to re-work your referral data to get an accurate picture.
Then, go for it. Build that custom attribution model. In Google Analytics navigate to Conversions>Attribution>Model Comparison. You’ll want to build your new model there by clicking on Select Model (above the table) and choosing “Create New Custom Model.” I suggest you build your first model based on the Position Model, and change your percentages to give much less credit to first click, significantly more to your last click, and spread the remaining over the middle.
A good place to start is from Avinash Kaushik’s Market Motive attribution model, which you can find in the Google Analytics custom model gallery. To start out, you’ll want to compare your time delay model to the custom model, but then use your knowledge of real-world customer behavior to inform your tests and changes. This is where you’ll likely spend the rest of your analytical days, building and tweaking attribution models to understand where your customers come from and where you can invest more resources to get greater ROI.
* * *
Using Analytics out of the box can give you a high-level understanding of where your traffic comes from and what your users do when they interact with your site. But to get a true picture of traffic behavior and your conversion funnel, you need to test and customize your models until they accurately reflect your customer journey and how your prospects think.