Many marketers claim to know how to measure sales by discrete channel, but until we can crawl inside a customer’s mind we’ll have to make some guesses in terms of correctly attributing sales. It’s possible for our guesses to become more educated and more accurate; it just takes some hard work
Sales attribution has gotten more difficult with the addition of sales channels (e.g., e-commerce, mobile, affiliates) and marketing vehicles (e.g., email, search, and DRTV). Attribution is further complicated by internal turf wars where multiple groups want—and need—to claim credit for sales. Organizational silos work against accurate sales attribution when data isn’t shared. The more complex and lengthy the buying cycle, the more difficult it is to develop defendable sales attribution. The list of obstacles goes on.
Why fight these complications? Because sales attribution is incredibly useful for:
- More efficient resource allocation to achieve cost-per-sale and sales quantity goals
- Determining true campaign costs and quantifying effectiveness
- Rank ordering media cost on a per-sale basis
- Informing where to spend the next budget dollar most effectively
This is exactly what direct marketers live for: data-driven efficiency.
Making sales attribution work
Marketers typically use one of a few primary approaches to attribution to accurately depict and credit their sales and marketing programs.
A rules-based approach gives credit to (usually) the last touch or (sometimes) the first touch. So, for example, a sale to someone who initially responded to a direct mail lead generation campaign might either be credited entirely to direct mail or entirely to the representative who concluded the sale. While relatively easy to do, the accuracy of this simple approach is difficult to defend, particularly with longer sales cycles and multiple marketing vehicles.
A more sophisticated approach to the rules-based methodology takes into account the concept of “contribution”—and explicitly identifies all vehicles and channels that played a part in generating the sale. This approach establishes a defendable and much-sought-after “best” or strongest assignment of primary sales credit while acknowledging other vehicles and channels involved in the sale. The contribution approach can be further enhanced by assigning weights or proportional sales credit to the influencing vehicles.
The most advanced approach to attribution is statistical modeling. Models are data-driven—but some art is still involved in tuning models to reflect reality. The statistical modeling approach is often used when defendable ROI impact is sought in the absence of closely associated response data, e.g. brand advertising.
Here are a few best practices to follow when you’re ready to move to a statistical model for sales attribution.
1. Start with actual sales, not marketing channels. Because organizations love their silos, marketers tend to think about channels first. How many prospects did we approach and how many responded? The more accurate approach is to look at sales and work backward. Start from the bottom of the funnel and move up. This helps eliminate biases. Ask, “What steps did each buyer go through?”
2. Associate directly attributed sales to the responsible campaign. For this you’ll need to have thought through the possibilities ahead of time. Each campaign and each marketing touch will need its own unique code, direct phone number, custom URL, etc. Digital channels are usually great for this. Other channels provide their own challenges.
3. Use a control group methodology to assign a sales lift to specific campaigns. If you can’t assign sales directly, keep one (random) group in your database separated out from your campaign. Then you can readily measure the incremental lift in sales associated with your campaign versus the “natural” rate of sales experienced in the control group.
4. Track touches over time for match-back and allocation analysis. A clean, de-duped database is critical for ensuring that sales can be matched back to the correct prospect. Another key piece of the puzzle is to maintain data on each campaign so all variables can be measured. Then you’re ready for advanced reporting, key insights, and modeling for the future.
5. Establish business rules for future attribution. To build a statistical model, you and your team will need to agree on what gets credit, when, and how much based on the data at hand. Naturally, as the number of marketing and sales channels increase, the complexity of the rules increase. (Then, of course, it’s important to test your model.)
Best practices in sales attribution require that you map all customer touches throughout the buying cycle and align your customer touch strategy with each point. Measurement and tracking are essential—and it’s equally important to maintain the data in a unified system. Otherwise you’ll be left with more questions than answers.
With these best practices, when the data comes in you’ll be able to manage against clear, defined metrics all parties have agreed to. Then you can optimize each channel. Every time you go out to your market you’ll learn more. This process of continuous learning is the best way to build a streamlined selling machine through marketing.
Tom Reid is executive director at Hacker Group.
