Welcome to the third article in CaliberMind’s DIY Attribution series! Today we’re focused on the tools of the trade, including the pros and cons of each of your options. But before we dive into the tools you’ll need, let’s talk about the level of expertise and the time commitment that goes into attribution.
Data Modeling Skills & Then Some
The hardest thing about getting an attribution model right is understanding what people want to use it for. The purpose or goal should always drive how we build attribution. This means having someone on the team who can ask questions of the CMO until they understand what the CMO is envisioning presenting in the board room as well as understanding practical applications for the broader marketing team.
The person who understands how attribution will be used and what that means for how the raw data will be incorporated may not be the same person who:
- Figures out how to connect all of the necessary data sources
- Models the data appropriately
- Maintains the model any time a new tool is purchased
- Designs the reports in a visual layer
We’ll discuss what can and can’t be accomplished using marketing automation and your CRM. If you want a full-scale solution, you’ll likely need a team with different skill sets to connect the data sources, normalize the information, create a data map that makes it so people and accounts can be discovered and connected from any system, and tackle complex language and modeling. This will take a data warehouse or data lake with ETL connections to your key marketing and sales tools (and possibly product).
Finally, there’s the person who will be taking that data and putting it into a visual layer and refining the look and feel until your CMO (and broader team) are happy. This could mean dumping calculated data into Excel and building pivots and charts or it could mean building a connection to Looker, PowerBI, Tableau, or some other BI tool.
Remember, any time you purchase a new marketing tool, the process has to be repeated to make the new data usable. You’ll need access to the same resources regularly. Initial setup has taken many skilled people a year or more, and each time a tool is changed, you’re likely looking anywhere from two weeks to two months depending on how accessible your resources are. If you don’t have someone who can do all the things (this is a unicorn, good luck finding them!) on your marketing team, you’ll be relying on an outside team with an independent priority list.
The Toolset for “Close Enough” Attribution
If you’re reading this article, you’ve probably already hit the limits of your CRM’s attribution capabilities. CRMs can provide you with multi-touch attribution, but it’s important to understand that you’re limited to the data housed in your CRM and that it’s not practical to push everything you want to see in an attribution model into that CRM. CRMs are protective of storage and processing time – and there are a lot of limitations around joins. Trend analysis and cohorting data are nearly impossible.
Your CRM attribution is also likely restricted to campaign touches. We’ve seen mad scientist operations professionals build campaigns and replicate sales tasks and even web data as campaign members, but we don’t recommend this approach. Building virtual campaigns in a database to house this information saves you a lot of data stored in your CRM.
If you want to understand how campaign data influences reports and you’re willing to do some fancy logic in your CRM to automate opportunity contact role creation on opportunities, you may be able to get a fair understanding of how marketing campaigns perform. This is great if you want to answer questions like whether or not prospects are engaging with your tradeshow investments or webinars. It’s not great if your CMO wants to present a slide showing the percentage breakdown of each department’s contribution to pipeline and bookings.
It’s also not great if you want to understand how website content or email marketing influences opportunities without blowing up how much “credit” marketing gets for pipeline and bookings.
If the extent of what you’re trying to achieve is campaign optimization, you can get a long way using multi-touch attribution or influence reporting in CRM. Where this option falls short is when your executive team wants a way to calculate a return on marketing investment model or demonstrate how each go-to-market team is contributing to bookings.
The Toolset Options for Multi-Touch Attribution For ROI Calculations
A data warehouse is another option for building an attribution model, as it provides a central repository for all the data needed to perform attribution analysis. With connections to core data repositories, such as marketing automation platforms, CRM systems, and web analytics tools, a data warehouse can be a powerful tool for analyzing marketing performance across channels and touch-points.
Another pro when using a data warehouse or data lake is the ability to consider cross-functional efforts leading to a booking. Whether we marketers like it or not, we’re ultimately judged by bookings attainment. We can’t control much beyond lead creation, but a company’s health isn’t determined by lead volume. A company’s health is determined by a business’s ability to win deals. Because sales plays a critical role in closing opportunities in B2B SaaS, we need to factor in their efforts before calculating ROI.
Still not bought into this approach?
Think of the most common way to think of spend efficiency: CAC or Customer Acquisition Cost. This is a formula that totals marketing and sales spend (including salaries and commissions) divided by the number of customers acquired. Finance takes the expenses of both sales and marketing, and it’s logical that we must consider what sales is contributing when building a model to estimate ROI.
However, building a data warehouse can be a complex and time-consuming process that requires specialized skills and resources. In addition, maintaining and updating a data warehouse can be an ongoing challenge, as new data sources and business requirements emerge. As such, it’s important to carefully consider the cost and complexity of building a data warehouse before embarking on this approach.
There are tools, like CaliberMind, that are structured as a customer data platform (essentially a data warehouse with all of the logic necessary to connect disparate data sources) and can be implemented quickly. Our average implementation is about eight weeks, with reports delivered in the first two weeks and time for quality checks built in to ensure the logic is customized to your organization.
Some other benefits to purchasing a tool purpose-built for marketing analytics:
- The ability to normalize data and push it back into your source systems
- Deduplication logic that can also be used to cleanse your CRM
- Additional reporting capabilities like engagement scoring at the account level and funnel analysis
- The ability to trend and cohort data natively
When you purchase a specialized product, you’re also purchasing years of learnings from implementing attribution across many companies. Data modeling is rife with missteps and pitfalls, and companies that have experience successfully managing multi-year customers have proven their models and refined them over time. You’re paying them for a product that is ready to go – not to learn how to start from scratch.
Ultimately, the choice of technology for building an attribution model will depend on a range of factors, including the size and complexity of your marketing operations, the availability of data sources, and the skill sets of your team. By carefully evaluating the pros and cons of each option, you can select the technology that best meets your needs and helps you gain a better understanding of the impact of your marketing efforts on revenue growth.
Tools to Avoid
There are technologies that can’t be used to build a multi-touch attribution model. One example is Google Analytics. While Google Analytics is a powerful tool for web analytics and reporting, it is not designed to track multi-touch attribution. It can track the first and last touch, but it does not provide visibility into the touches in between.
Another example is social media platforms like Facebook, LinkedIn, and Twitter. While these platforms provide powerful targeting and analytics capabilities, they are not designed to track multi-touch attribution. They can track clicks and conversions, but they do not provide visibility into the entire customer journey.
Attribution Isn’t a Set It & Forget It Kind of Thing
Once you have built your attribution model, it’s important to test it and refine it over time. This will involve analyzing your results and making adjustments as needed. You may also want to consider A/B testing different attribution models to see which one works best for your business.
As your business grows, your models will need to be revisited. Getting cross-functional buy-in is an ongoing process, so socializing your model and ensuring new business leaders are very familiar with the logic and process you use is as critical as maintaining new connections when tools are swapped out in your marketing tech stack.
Building your own attribution model can be a challenging task, but it can also be a great way to gain deeper insights into your marketing campaigns. Whether you choose to hire an expert or build your own model, it’s important to keep in mind the pros and cons of each option. With the right tools and skills, you can create a powerful attribution model that will help you make data-driven decisions and improve your marketing ROI.