Wouldn’t you like to know which marketing channels your prospects traveled on their way to becoming your customers? That’s the purpose of attribution modeling, the process of identifying which touch points across marketing channels that contribute to a conversion, and then assigning a value to each channel to ultimately calculate the return on your marketing investment (ROMI).
The challenge, of course, is that B2B customers rarely take a linear path to your door. Consider this likely scenario.
Let’s say you’re a start-up going to SXSW for the first time and you really need to make a splash. Maybe you’re looking for a stellar but affordable local exhibition company to create your graphics and in-booth experience. Where do you begin?
With a web search, of course. Using your search engine of choice, you enter “great exhibition company in Texas” and get back many options, both in the organic listings and in the sponsored ads.
You click on an organic listing for “Rockstar Texas Exhibition Company,” go to the website to check it out, then decide to hit Facebook and Twitter, and Google+ to check out some customer reviews. Feeling hungry, next you search for “best BBQ in Austin” and get hit with a retargeting display ad for the same “Rockstar” company you clicked on in the first place. So you decide to click on the ad, which brings you back to the company's website, where you decide to purchase an exhibition package.
The Rise of “Search Sessions”
That right there is what’s called a “search session.” You went through multiple touch points before deciding to hire the exhibition company. Which one gets the credit for making the purchase happen? Rockstar would like to know how each channel is performing so they can measure the effectiveness of their marketing investment and figure out how to make the most of their marketing budget.
Let’s take a look at each of the popular types of models and see how much value is attributed to each step. This way we can hone in on some pros and cons.
This attribution model gives credit to the last channel before conversion, which would be the retargeting ad in our example. Last non-direct click gives credit to the second to last channel before conversion, which would be the Rockstar Google+ page. Last-click AdWords gives credit to the most recent AdWords click before conversion. (See more here.)
- Pros: Last-click attribution has low data requirements because only one touch point is tracked, and the set-up is less technical, particularly with mobile because many networks only record clicks and not impressions. It’s also the best way to get an apples-to-apples comparison across all channels and campaigns.
- Cons: Last-click attribution is biased toward middle of funnel (MOFU) and bottom of funnel (BOFU) tactics. It cannot tell you how a prospect found you or how many times they interacted with your brand during a search session, which may throw you off course with budgeting as you move dollars away from channels that are helping your prospects convert.
This attribution model gives credit to the first channel before conversion in a search session. In the Rockstar example, this would give the Rockstar organic listing all the credit, suggesting more dollars should be moved to SEO.
- Pros: First-click attribution will provide insight into how prospects found you, is easy to set up and, like last click attribution, it does not require a ton of data.
- Cons: It takes more than one search, click, or like for a B2B buyer to find and consider a solution, and first-click attribution doesn't account for all of the subsequent touches that take place during their journey. The first click model also places more value (and therefore, more budget) on channels used in the early stages of the buying cycle rather than those used in the middle or toward the end, when a prospect is closer to making a purchase.
This attribution model allows you to create a hybrid of the of the last-click and first-click models and give a desired percentage of credit to each touch point. In the Rockstar example, we would give 40% credit to the organic listing, 40% credit to the retargeting ad, and equal credit to the middle touches on Facebook, Twitter, and Google+.
- Pros: This is a great model to use if you value not only the touch points that introduce customers to your brand, but also the final clicks that result in conversion.
- Cons: If your data isn't clean, or if you have little experience attributing values to each touch point, this model can steer you in a bad direction—fast.
This attribution model gives equal credit to all channels before a conversion. In the Rockstar example, that means that organic search, Facebook, Twitter, Google+, and the retargeting ad would all get equal points for bringing in the sale.
- Pros: Linear attribution can be useful if your marketing campaigns are designed to maintain contact and awareness with prospects through the entire customer journey. For example, if you're running a brand campaign and each touch point is equally important during the consideration phase, linear attribution will help you visualize this process.
- Cons: Linear attribution runs the risk of duplicating marketing efforts because you cannot be sure which touch had the biggest impact, which means you may end up investing more in a particular channel than you should.
This attribution model uses an algorithm to give the most credit to the channel closest to the conversion and increasingly less to the rest of the channels. This means that as the Rockstar prospect gets closer and closer to purchase, whatever channel they’re in gets increasingly more credit.
- Pros: Time decay attribution can be customized to map to your sales cycle, particularly those sales cycles that have shorter consideration phases. If you are not doing a custom model and doing a lot of testing with your data, time decay is a feasible model that more truly imitates prospect behavior, because prospects become more acquainted with your brand as they have had more time to consider purchasing your goods or services.
- Cons: Time decay attribution discounts initial touch points that may have had the most influence on the final conversion, such as a post on Facebook or the beginning of a search session.
This is the “Holy Grail” of attribution models! You can mold your model around more specific business questions and objectives and compare your custom model and other default models side-by-side. Here is an example of how one would look for Rockstar:
- Pros: With this model you can use the linear, first-click, last-click, time decay, and position-based attribution models as your baseline, and then layer in other factors important for your business. The best part is that it is personalized for your website and your business and can be optimized over time.
- Cons: Do not attempt to use custom attribution without first doing some controlled tests on your data and campaigns to see which variables and channels actually make a difference in your conversion values. You should get to know the other models very well before jumping into this arena.
As you can see, there are many attribution models out there and the only way you’re going to find one that is right for your business is to test and optimize each one. Use some dummy data sets to test different campaigns and channels. There is no “perfect” model out there because your business, industry, and data will always be changing. Don’t let this scare you. Let this empower you to start looking into a good measurement plan and get started on your way to be a solid competitor in your space.
Ready to get started? See how the Yesler team approaches marketing analytics and data analysis.