Rethinking today’s attribution problem in digital marketing
Attribution is the science behind assigning values to individual touch points throughout a customer’s decision journey. It’s not only a key driver in how we currently optimize marketing campaigns, but attribution is also used for media mix modeling and developing media budgets.
Herein lies the problem: marketers tend to look at attribution within our own channel silo, so it’s not easy to understand the full picture of a multi-channel environment.
As search continues to evolve in form and function, it’s raising some important questions around how we think about and use attribution models — and is putting some of our long-held practices to the test.
Do you have an attribution problem?
Attribution modeling has seen incredible progress over the years, offering increasingly better solutions to track touch points throughout a conversion. While progress has been made, the attribution models that most companies are using today offer a partial solution that tracks only online conversions and channels.
Most companies are just beginning to scratch the surface of multi-channel and cross-device attribution. Technology companies are still piecing together the components needed to provide a full solution to fill in the gaps in current attribution models. It leaves us to question, do we have an attribution problem that we might not be aware of?
The simple way to answer the question is to look at your analytics data and see if a significant portion of overall conversions have a long, multi-step path. Or in other words, are your customers engaging across multiple channels? If the answer is yes, a significant portion of your conversions come from paths that contain 2+ steps, then you most likely have an attribution problem.
Here is how you can see this for yourself: In Google Analytics, navigate to Conversions > Multi-Channel Funnels Report, and then look at the Path Length report. This report will provide you with a simple breakdown of the quantity of paths. You should also view the Top Conversion Paths report to understand which channels play a role at each step of the consumer’s digital journey.
I do want to acknowledge that this report is limited to just your digital channels, unless you’re pulling in additional conversion data from offline channels into analytics.
Mind the attribution gap
In June, Aaron Levy wrote an excellent post on the best attribution model for search. In it, he describes the top rules-based attribution models, along with the high-level pros and cons of each model.
I’m going to dig a little deeper into the topic, and I’m going to start by discussing the two major attribution gaps you need to be aware of when it comes to attribution models.
The first gap in attribution is in understanding the impact of online channels’ ability to drive offline revenue. Most analytics platforms and their attribution models are only providing a partial solution by looking at the digital conversions and revenue, not including the online impact to offline sales/leads/in-store visits.
Of the two major gaps in attribution, this is one of the easiest gaps to solve with moderate accuracy based on extrapolating information from available data sources to link online activity to offline user behavior.
The easy solution that many businesses take is to use some sort of unique promotion code by channel to tie attribution together. This gives consumers a choice in where they can convert and allows you to track the conversion.
Google is working on solving this attribution problem with their “In-Store Visits” metric, which uses a number of signals to gauge online-to-offline impact, including Google Maps data, GPS, WiFi, visitor queries and data from over a million opted-in users, which is then used to create store visit estimates.
Analytics and advertising platforms are leveraging data from cell phones such as WiFi, GPS and beacons tied to in-app networks. Even mobile forms of payment like mobile wallets will help to answer the online-to-offline attribution question. While these methods aren’t a perfect solution because they don’t track a specific user’s behavior, they are a start in the right direction to solving the online-to-offline gap.
The second and most difficult gap in attribution is in measuring consumer behaviors and interactions across screens and devices. Are we tracking consumers across devices and across channels? Can we?
This is an attempt to answer the questions, “How do consumers engage with a brand across devices?,” “Which channels are they engaging with on each device?” and “On which device do they finally convert?”
This is one of the biggest attribution challenges because consumers use a multitude of screens (TV, desktop, tablets, smartphone and more) throughout their journey. As consumers switch across devices, it’s difficult — nay, almost impossible — to maintain the unique IDs to track the customer journey both offline (through TV viewing) and online. This is going to be the most difficult gap to close due to the complexity needed to track unique users across multiple screens.
Today, advertisers attempt to bridge the gap by looking at data and finding correlations between interactions across screens. There are a few large companies who can close this gap through logged-in states across multiple screens and devices (think Apple, Facebook, Comcast), though potentially not across all of the screens and devices we use regularly.
The cross-screen, multi-channel attribution will only get more difficult as search powers experiences away from the search box, such as voice search through personal digital assistants and Virtual Reality/Augmented Reality technology.
Regardless of the attribution model chosen and the gaps that exist across all attribution models today, the purpose is still the same: to understand the value each channel brings to the marketing mix in moving customers along their decision journey.
The goal we should be focusing on as marketers is understanding how to integrate search into the journey and create more holistic campaigns that focus on the consumer. Tallying touch points from a rules-based model may be detracting from the higher goal of creating deeper, lasting customer relationships.
First- & last-click see their last days
In my own experience, many companies employ either first-click or last-click attribution models. First-click (or first-touch) attribution models give 100 percent of the credit for a conversion to the consumer’s first touch point, whereas last-click attribution models assign conversion credit to the last touch point leading up to a conversion.
It’s easy to scrutinize either of these popular models because they focus solely on either the top of the funnel (first-click) or the bottom of the funnel (last-click). Indeed, as a result of their disregard for all other marketing activities, both first-click and last-click attribution models are heading toward their last days.
Using the last-click model means mean that you are either ignoring early, top-of-funnel activity and instead focusing on bottom-of-funnel elements like branded search and remarketing (which tend to drive the final conversion). Without giving value to the top-of-funnel channels, sooner or later, your remarketing efforts will dry up.
First-click attribution comes with similar problems — by giving all the credit to the first touch point, you obscure the true value of critical middle- and bottom-of-funnel efforts in moving the customer through the buyer journey and closing the sale.
Other attribution models, such as time-decay and metric-driven, offer more sophisticated modeling but still end up assigning arbitrary values and leaving gaps when customers switch between channels, especially when going from online to offline.
Even within our own channel, search is no longer about singular conversions. Keywords are now being evaluated in a much larger context, not just for immediate conversions but how they assist throughout the consumer decision journey. They are helping consumers explore, research, compare, locate, purchase and follow up.
Search has seen rapid growth in both form and function since its inception less than 30 years ago. In form, it’s broken out of the text box, appearing on our phones, in our cars, at home on our speakers, in our TV remotes and gaming systems. It’s in bed on our tablets and phones. Wherever we go, search is with us across devices.
In function, we’ve started relying on search as inter-communicative partners. Search has gone beyond simple voice input and is evolving to understand user intent and behaviors through available data to help consumers take action. Through search, we can order a pizza, rent a car, buy movie tickets and compare insurance quotes — all within the results pages.
How could search, or any digital channel for that matter, possibly still be graded on one click prior to a conversion? It simply can’t.
The future of campaigns
Rather than being a siloed line item in the marketing mix, search has proven itself to be an integrated component of a much larger machine focused on lasting relationships or CLVs (customer lifetime values).
Today’s marketers are adopting a new approach that tears down the siloed walls of search, social, display and offline channels to create one holistic brand experience. Holistic campaigns support deeper, more meaningful customer relationships and distinguish themselves with the following factors:
Getting personal with digital fingerprints
Creating highly personalized campaigns is an extremely effective way to strengthen your customer relationships. There are many new tools in search that allow you to personalize the search experience, such as remarketing or customer match. But before you launch individual campaigns, you should first think holistically about how a campaign will integrate with other campaigns.
Each consumer has a unique digital fingerprint, which is an evolved concept from the traditional “digital footprint” that centers around following a user’s online trail. A digital fingerprint signifies much more: a user’s unique online preferences as they embark on highly individualized journeys, “depending on cost of failure, frequency, cost and complexity of the task, and the type of shopper.” (Bing Ads Customer Decision Journey report).
Each consumer displays highly personal patterns and preferences throughout the relationship-building process. They may rely more heavily on email, spend more time on social, prefer tablets and so forth.
Your challenge will be to create a holistic brand experience that consumers can tap into based on their unique digital fingerprints. Although the three consumers below have unique digital fingerprints, they should all experience a unified, holistic brand.
Rethinking your future campaigns
As you rethink your future search campaigns with a new, holistic approach, we recommend the following:
- Carefully select an attribution model that will support — and not work against — holistic marketing efforts. Refrain from using oversimplified models, such as either first or last click, which will not sustain long-term multi-channel campaigns.
- Evaluate your keywords based on more than the final conversion metric. Look at assists to make sure that you’re supporting customers throughout the entire consumer decision journey. Look at other metrics based on where the keyword fits within the journey, such as time on site, page views, product page views and other customer loyalty indicators.
- Include keywords that support customers throughout each stage of the consumer decision journey as they engage with search as an inter-communicative partner
- Include campaign-specific keywords within your search campaigns so that consumers can easily find you after seeing a TV ad, remembering a billboard, hearing your jingle or seeing your logo.
- In the case of tracking online to offline or vice versa, include unique promotional codes to tie purchases back to a unique marketing channel.
- Be aware of your consumers’ unique digital fingerprints and how your customers will experience your brand across different channels at different times with different connectivity points.
Some opinions expressed in this article may be those of a guest author and not necessarily Search Engine Land. Staff authors are listed here.
Originally published at searchengineland.com on October 12, 2016.