Retail marketers have debated the end of last touch attribution for over a decade now. In today’s omnichannel marketing world, where consumers shop across a variety of devices and channels, attributing a sale wholly to the last interaction before purchase is outdated and inefficient. That’s because you can overly emphasize low-funnel activities at the risk of neglecting efforts to fill the top of the funnel.
The customer journey has grown increasingly complex, so optimizing for a single action or channel ignores a much bigger picture of how consumers engage with a retailer’s brand messaging on the way to purchase.
A customer, for example, may discover a new product on Pinterest, rediscover that product through display ad retargeting, research the product via Amazon, and finally make the purchase after receiving a promotional email. This journey can occur across desktop, mobile and voice assistants. Valuing only the last touch, the email click ignores the very important role each marketing activity had in driving the sale and limits marketers’ ability to develop greater cross-channel cohesion.
While many marketers recognize the omnichannel nature of online retail and the limitations of last touch, few have made the switch to a more accurate, multi-touch attribution model. The reasons for this are varied. Changing marketing attribution can be time-consuming and requires lockstep collaboration across departments. Gaining buy-in from executives can also prove challenging because it is difficult to quantify the revenue and ROI payoff.
The end of last touch?
Despite these challenges and marketers’ inertia when it comes to last touch attribution, the time is now to audit your attribution model. You could gain an advantage over competitors who still use last touch by identifying new opportunities to grow your customer base and develop more impactful cross-channel marketing initiatives.
As competition grows on paid search, social and display marketing channels, and costs creep steadily higher, you will need to find a way to not simply spend more, but spend smarter. Multi-touch attribution provides a fresh perspective on performance and can point out which channels are contributing most to the sale but may be undervalued in terms of a marketing budget. A truly granular approach to attribution can even identify a specific ad creative that is outperforming other ads in your marketing mix.
Major advertising platforms like Google have also hinted that they will be moving away from a last touch attribution model. While last touch may not disappear entirely from Google Ads, Google will likely emphasize its multi-touch attribution reporting, and in particular Data-Driven Attribution (DDA), which uses algorithmic attribution to credit sales across Google’s ad properties.
All of these competitive and technological forces place retail marketers at a crossroads. Should you maintain the simplistic single touch approach or should you explore new attribution models?
Changing attribution models, particularly from something like last touch to one of the many multi-touch attribution models (U-shaped, W-Shaped, time-decay, algorithmic, etc.) is no easy task. That’s why it’s critical to evaluate whether this switch makes sense for your business. Before making an attribution change, ask yourself these four questions:
1. How many marketing channels do you advertise on?
Multi-touch attribution is most valuable for marketers who use a variety of channels to nurture consumers towards a sale. If you spend marketing dollars across email, Facebook, Google Shopping, paid search and Amazon, for instance, it may make sense to track attribution across all of these channels. That will help you identify the white space in your marketing efforts.
If, on the other hand, 90 percent of your marketing spend goes towards Facebook, for example, it may not make sense to invest in multi-touch attribution at this point. In the future, though, as marketing initiatives expand to new channels and grow in sophistication, you may want to consider evaluating various multi-touch models.
2. How many touches typically occur before the sale?
If you have a long buying cycle where every touch is important, an evenly weighted, linear attribution model or custom algorithmic-based model may make the most sense. A shorter sales cycle, on the other hand, may lead you towards an attribution model that weighs touches later in the sales cycle more heavily, such as a time decay model. If very few touches occur before the average sale, investing in a multi-touch attribution model may not make sense.
A simple way you can get a better view of your customer journey is by using the Multi-Channel Funnels reports in Google Analytics. In these reports, Google tracks how different referrals and searches contribute to a sale.
Google provides four different types of Multi-Channel Funnels reports. The Top Conversion Paths report shows how customers interacted with varying channels of marketing and searches before purchase. The Time Lag and Path Length reports show how long in days and interactions it took for users to become customers. And lastly, the Assisted Conversions report shows how many sales each channel initiated, assisted and completed, as well as the value of those conversions.
3. Can your business support a transition period to accurately report year-over-year metrics?
When you implement a new attribution model and use that new model as your source of truth, you lose the ability to make year-over-year comparisons with your data for the first year. Any sales data tracked with the last touch model, for example, cannot be compared to time decay attribution data.
If comparing year-over-year numbers is critical to your business, account for a year-long transition period. Plan to run your new attribution model alongside the legacy model for that first year. This allows for better data comparisons and doesn’t throw a wrench in year-over-year reporting.
4. Is competition making your performance marketing strategies less impactful?
Perhaps one of the biggest reasons to adopt a new attribution model is to discover the gaps in your marketing initiatives. If a certain channel or performance marketing campaign is no longer driving the same ROAS due to increased competition, it may be time to look at your data in a new way. A multi-touch attribution model can shine a light on a channel or campaign that seems to be underperforming in the last touch model but is contributing to sales when viewed in another way.
An easy way to view your campaigns through different attribution methods is by using the Google Analytics attribution model comparison tool, found under the “Conversions” tab in Google Analytics.
For example, in the below screenshot, you can see that performance for a paid search varies significantly when viewed through the lens of last touch, first touch, and position based (also known as U-Shaped) models. With the last touch model, it appears that paid search campaigns only contributed one conversion, at a costly CPA of $6,985.33. But first touch and positioned based models paint a completely different picture, with 69 and 34.69 conversions respectively.
In the last touch approach, paid search would be written off as a waste of time and ad spend, but in reality, the channel is helping fill the funnel by engaging shoppers early in their journey and nurturing them towards a purchase.
Attribution modeling provides a full-funnel advantage
Regardless of which attribution model you choose, it’s critical to start viewing performance marketing data in ways that more accurately reflect how your audience shops today. That means valuing the many channels and devices shoppers interact with and gaining a better grasp with how different marketing initiatives perform at different stages of the funnel. Investing in an attribution model that reflects these nuances will empower your organization to spend smarter than its competitors and identify previously overlooked initiatives that can grow the marketing funnel and your market share.
For retail marketers with a robust marketing strategy and an omnichannel sales approach, any attribution model is better than last touch.
Opinions expressed in this article are those of the guest author and not necessarily Marketing Land. Staff authors are listed here.