By Ryan Koonce
Multi-touch attribution is the practice of assigning credit to marketing touchpoints in proportion to their impact in driving a desired conversion outcome. The operative term here is “proportion.” Multi-touch attribution models collect data from all touches in the buyer’s journey and assign proportionate credit based on the attribution weightings of a model, thereby accurately reflecting their impact (or not) in generating a sale.
For example, if a prospect or visitor comes to your website through a search ad, converts to a lead on a newsletter, engages in person with a sales person, and revisits your website through a remarketing ad to make a purchase for $100, multi-touch attribution (using a linear model for illustrative purposes) would assign $25 in proportionate revenue credit to each of the four touches. And depending on your view of the impact of the recency of the touch, multi-touch attribution models come in basic flavors to easily fit your strategy.
Unlike multi-channel attribution, multi-touch attribution can track any marketing touch with a deep level of granularity, such as online channels (any source, medium, campaign, etc.), events (any sales touch), email campaigns, and even direct mail, mobile devices and television.
Remember the earlier word, “proportionate”? The problem is single-touch attribution allocates a disproportionate amount of revenue credit to either the first or last touch. If your customers’ purchase path is straightforward, single touch attribution is an adequate method of informing budget allocation using tools like Google Analytics, CRM or ecommerce. However, like most modern B2B and B2C businesses, you’re probably marketing through multiple paid platforms and marketing channels including offline, single-touch attribution misrepresents ROI by giving 100% revenue credit to either the first or last, woefully understating the impact of other touches and channels.
For example, if a prospect or visitor comes to your website through a search ad, converts to a lead on a newsletter, engages in person with a sales person, and revisits your website through a remarketing ad to make a purchase for $100, multi-touch attribution (using a linear model for illustrative purposes) would assign $15 in proportionate revenue credit to each of the six touches. And depending on your view of the impact of the recency of the touch, multi-touch attribution models come in basic flavors to easily fit your strategy.
Further, single-touch attribution data from ad-specific ad sources such as Google Analytics and Facebook Insights often double-count revenue with each source claiming 100% revenue credit to a sale. With this “attribution anarchy,” marketers can’t make data-informed decisions as to which channels and campaigns are worth spending more dollars on and which ones are wasting dollars.
While single-touch attribution only gives credit to one marketing touchpoint, multi-touch attribution models assume all touchpoints play some role in driving a conversion such as a lead, pipeline or a sale. Here we’ll look at the most popular multi-touch attribution models and how you can choose the one that’s best for your marketing.
A linear or “impartial” model is the baseline for multi-touch attribution models. It assigns an equal percentage of revenue credit to each touch regardless of its recency in the buying journey. Hence, using the earlier example, each touchpoint receives an equal $15 in revenue credit, totaling $100.
If you don’t have a strong view of the value of when the touch took place in the buying journey, then a linear model is an excellent starting point. Don’t try to boil the ocean! The good news is with a modern multi-touch attribution tool you can easily toggle between attribution models to see the differences and start learning.
A time-decay model assigns revenue credit to each touch based in accordance to its recency in the buying journey. Whereas a linear model gives equal credit, a time-decay multi-touch attribution model says the closer the touch was to the sale or conversion event, the more influential it was. Using the same buyer journey, you can see the more recent touches receive more credit than the older ones.
If you believe strongly (and have some supporting data) that more credit should be allocated to touches that nudged prospect closer to a sale vs touches that brought the lead or buyer in, then a time-decay multi-touch attribution model is the way to go.
A position-based (or U or W-shaped) model assigns 40% revenue credit to the first and last touch prior to the conversion event or sale, with the remaining 20% distributed equally among the middle touches. If there is only one touch before a conversion event or sale, then a position-based multi-touch attribution model would give 100% revenue credit to the touch. Similarly, if there is only a first and last touch, they each receive 50%.
Let’s look at our example to see how these models work:
Note: A W-shaped model works the same as a U-shaped model except it allocates 30% revenue credit each to the first, key middle, and last touch, while distributing equally the remaining 10% among other middle touches
If your business places more value on the touch that brought in the initial prospect and the touch that converted them compared to the middle of the buyer journey touches, then you should use a position-based multi-touch attribution model to optimize your campaigns and channels.
Custom machine learning modeling is the most advanced approach to multi-touch attribution. A custom model uses a machine learning algorithm to assign revenue credits to touches. Rather than using a static model with user-determined weightings as in the above examples, an algorithm is derived using historical touch and conversion data. As you can see in the example below, data-driven modeling applies a unique algorithm to allocate credits.
It is for this reason you should not expend the effort with it until after you’ve gone through at least one or two full path buying cycles using one or more of the static models. If you’re business is direct-to-consumer with a short purchase path, this means a matter of a few weeks, buy if you’re dealing with longer B2B buyer journeys I would learn from the static models for a good six or more months depending on your sales cycle to determine if there is ample opportunity to further optimize your marketing mix.
There’s no perfect science for choosing the right multi-touch attribution model. As the use of multi-touch attribution to optimize budgets and channels matures, marketers will run several models at one time to see which one is right for their business. The good news is modern multi-touch attribution tools allow you to see and compare the results from each model.
Regardless of whether you’re a B2B or B2C marketer, multi-touch attribution is worth the effort if you’re on the hook for lead or pipeline generation and/or customer acquisition or purchase, AND you’re:
If you have a short and simple marketing funnel, a single-touch model may be good enough. But if you’re marketing on various channels, have many touchpoints and are nearing “attribution anarchy” with single-touch tracking, you’re ready to move to multi-touch attribution marketing. Now.
Because buyer journeys span many touchpoints and devices before converting, marketers need to understand which touchpoints a buyer (or buyers in the case of B2B account-based marketing) interacted with that resulted in a positive action. The goal is to understand where to focus spend, devoting funds to higher performing channels and campaigns and diverting dollars from those that were ineffective, making the change to multi-touch attribution marketing is as much a culture shift as it is a new technology. With the click of a mouse, it enables marketers to look at all channels and attribute the sale to those touches using various models. Marketers can look at user-level data (clicks, forms) their touches have on key conversion goals such as leads, signups and purchases. Unlike single-touch attribution, multi-touch attribution models enable marketers to better understand both the chronology and the type of interactions that preceded and influenced conversions, using the insight to optimize the conversion paths of buyers.
E-commerce, driven by multiple new channels, is a booming business. “The NRF expects that online…
I’m excited to announce new features that will enable marketers to integrate television advertising and…
The evidence that CMOs are on the hook for the magic growth number is overwhelming,…