SPLITIT LTD (Company) Objectives The Risk, Audit and Governance Committee (Committee) has been established by the board of directors (Board) of the Company pursuant to article 47 of the Company’s articles of association. The Committee’s primary objective is, in
The Science – and Art – of Marketing Attribution
When you spend money on advertising, paid social media, or even influencer marketing, how do you know whether it’s effective? Digital advertising platforms sell themselves heavily on your ability to track what happens. You might think that measuring effectiveness doesn’t create any challenges as a result.
In reality, measuring effectiveness doesn’t come quite so easily. A few simple scenarios come to mind:
- A shopper sees your paid display ad on a website. They click through to your product promotion, but they don’t make a purchase. Later, they remember what they found interesting, and they do a web search to find it.
- A shopper sees your brand on their Instagram feed, but they don’t click on the post. Later, they go directly to your website because your URL is easy to remember or guess.
- A shopper clicks on your paid search ad. Then they spend a lot of time exploring your site, visiting competitor sites in the same tab, using the back button, and following a complicated path before making a purchase.
- A shopper goes straight from your paid ad to placing an item in their cart on your site. They abandon their cart and then come back to it later because you sent them a reminder email.
In each of these scenarios, it’s hard to draw a straight line from cause (your ad) to effect (a completed purchase). As a result, it’s hard to measure which ads are working and which marketing campaigns are successful.
Marketers have known about this problem since the early days of digital advertising – and that’s where “marketing attribution” comes in. It’s an intimidating term for many e-commerce retailers.
Worse, it comes with many specialized terms that get very wonky very quickly if you’re not deep in the weeds of this particular area of marketing analytics. Attribution modeling, click attribution, touch attribution, linear attribution, position-based attribution — these are just the tip of the iceberg.
Have we scared you off yet? We have some good news, too. In what follows, we’ll explain what you need to know about attribution as an e-commerce marketer. We’ll explain all of the complex jargon and more importantly, what it means for you. We’ll help you understand what’s possible, and what you need to make it work.
And… don’t miss our downloadable guide to attribution dos and don’ts to help you start making better sense of YOUR marketing effectiveness.
What Is Marketing Attribution?
Marketing attribution is the practice of assessing the dollar value of each marketing touchpoint in your marketing mix. It goes beyond the typical Google Analytics data that measure your web traffic. Here’s a simplified illustration of how this works in practice.
If you know the direct revenue from each marketing channel, measuring the ROI is the easy part. But no one ever said e-commerce is easy!
The bigger challenge lies in figuring out how you actually assign a purchase to a touchpoint. In today’s world, shoppers encounter your brand and your offering across many touchpoints along your sales funnel before they make a purchase. Each customer journey is unique. There are two things you need to know about this challenge.
- Perfect attribution doesn’t exist. We just don’t have the tools or data to capture the realities of each shopper’s journey from awareness to purchase.
- Ultimately, it comes down to your attribution model. This term refers to the method you use to determine attribution. There’s no universal right or wrong answer. It’s a choice that you make based on what’s possible with the data and resources you have, and what gives you enough information to make decisions that improve your marketing effectiveness.
Types of Marketing Attribution
Let’s now take a look at the different types of attribution models. They will give you a better sense of your options when choosing an approach to attribution modeling.
- First-touch attribution: Assigns leads and purchases to the first click or first interaction. In other words, if your shopper saw your ad on Instagram featuring your newest products and signed up for an email, then purchased after receiving an email promotion several weeks later for a 20% discount on their first-time purchase, you would still attribute the revenue to Instagram. It helps you understand how you’re capturing new customers’ attention. It’s very simple to do, but it doesn’t create a very clear picture of cause and effect.
- Last-touch attribution: Assigns leads and purchases to the last interaction. In other words, with the example above, you would only attribute the revenue to the promotional email instead of Instagram. It helps you understand how you are converting potential interest into purchases. Last-touch attribution is also very simple to do, but it doesn’t create a very clear picture of how you acquire new customers.
- Multi-touch attribution: Assigns purchases to all of the interactions the shopper has with you along their conversion path. Let’s say a shopper first clicks on your paid ad in Google. Then you retarget them with an ad on a newspaper’s website. Then they subscribe to your email list and finally they make a purchase from a Facebook ad. All of these channels get counted in your attribution model. But how? That’s where the next set of terms comes in.
- Linear attribution: This form of multi-touch attribution assigns equal weight to each interaction. In the multi-touch example above, you would attribute 25% to each channel.
- Position-based attribution: This form of attribution puts more weight on the first interaction and the last interaction. Anything in the middle is divided equally. So, with the same example, you would attribute 40% to paid search as the first click, 40% to Facebook as the last click, and split the remaining 20% between retargeting display ads and email.
- Time decay measurement: This form of attribution also factors in the amount of time that passes. In our simple example of Instagram and email, if a new customer first clicks an Instagram ad, signs up for email, but then doesn’t make a purchase until receiving an email promo eight months later, you would only attribute a small percentage of the result to Instagram. Most of the credit would go to your email channel which triggered their last click.
Believe it or not, there are actually many other types of marketing attribution model you can use. The attribution platform Bizible describes as many as 11! You can even develop your own weighting based on your unique sales cycle and marketing mix, or based on the way that you validate your model statistically.
And that last point is very important – because it gets to the heart of how you choose the attribution model that’s right for you and the resources you need to do attribution well.
The Skills and Resources You Need
Using an inappropriate attribution model or misinterpreting your attribution analytics can cost you a lot of money. You risk making the wrong decisions about your marketing spend or even starting to lose sales revenue because you’ve been misled by the data. Statistics can be tricky if you don’t approach them in the right way or if your data isn’t as representative as you think. You can easily and inadvertently introduce what’s called “statistical bias.”
There are many kinds of statistical bias, which we won’t cover in this guide, but the issue is easy to understand with a few simple examples.
- If you ran your attribution analytics during a period of heavy sales (such as holiday shopping) or curtailed sales (such as the first few weeks of the COVID-19 pandemic), you would get a skewed picture of which marketing channels work best and when.
- If you do any non-digital advertising, such as broadcast or out-of-home, which larger e-commerce brands often do, you’re likely to place too much weight on digital channels because offline channels are so difficult to measure.
- If you have multiple sales touchpoints (such as your own e-commerce store, an Amazon storefront, or even a physical store), you can easily miss the customer journey from initial awareness to a final purchase.
- If your marketing creative isn’t connecting with your target customers, for example, on social media, you might mistakenly conclude that you should spend less on social media rather than spend more with better copy and graphics.
- If your shopper isn’t ready to make a purchase, they might remember an ad they saw but didn’t click, and then search for your company name when it’s time to make a purchase. Your paid search ad would be their first click attribution in this scenario. The purchase should be attributed to the earlier, more memorable ad, but your data will tell you otherwise.
Having access to the right expertise can help you avoid problems such as the wrong data, wrong model, or wrong inferences.
To do marketing attribution well, you need to be able to capture data at every step of the customer journey, from their first click attribution or other first interaction to their last click attribution or other last interaction, and everything in between.
Assuming you have a multichannel marketing strategy, you also need to connect the dots from one platform to another: paid search, display ads on ad networks, social media, and potentially more. These platforms don’t necessarily talk to each other.
You also need clarity on what you can and cannot capture with attribution data. New customers might find you through a paid search ad, but not make a purchase decision until they have read third-party product reviews. Data will tell part of the story but it won’t illuminate every step on their conversion path.
The common tools for marketing attribution all offer various ways to help you bring data together across various marketing channels, but they aren’t all created equal. Some of them also help you track data even when you can’t tie it to an individual customer record, but it also depends on the other kinds of shoppers and shopping cart data in your infrastructure. And don’t forget—you still actually have to implement the tracking technologies that your tool makes available. See more detail on attribution tools below.
You don’t necessarily need to have a data scientist on your marketing team, but you should use resources such as this guide to increase your data literacy as well as anyone who will make or influence decisions about your marketing mix. You can also download our guide to attribution dos and don’ts for more information about how to make it work for your company.
You should also strongly consider bringing in expert help, either independent consultants or any customer onboarding resources offered to you by the marketing analytics platform that you use.
An expert’s knowledge and experience will allow you to make better decisions more quickly. They can advise you on which attribution model is best given your business goals, marketing channels, customer journeys, and conversion paths. They can also assess what data you can actually capture between first touch and last touch as well as what you can do to close any critical data gaps. Finally, they can guide you past the common mistakes and pitfalls that you might encounter as a newcomer to marketing attribution.
Once you get the data right and ensure that you have access to the right skills, you also have to make sure you set up the right processes for reviewing data. Reviewing data sporadically won’t help you achieve better marketing outcomes or optimize your marketing efforts. Reacting to every blip or change will just create chaos.
First, make sure that you have processes for reviewing attribution analytics for specific marketing campaigns. This approach will allow you to do more of what’s working and do less of what’s not working, even in the middle of a campaign.
It’s a good idea to do this several times. Do an initial review of marketing campaigns at around the 25% point (i.e., after one month of a four-month campaign). If you see any red flags or signs that one channel is already strongly outperforming, you can adjust your marketing mix accordingly.
Similarly, as you accumulate more data or depending on the duration of your sales funnel, you can perform these so-called “in-flight” checks at the 50% and 75% point of your campaign.
For each in-flight check, the goal is to make any necessary changes to ensure that your campaign delivers on its business goals.
When a campaign ends, you should also carry out a post-mortem of your entire campaign. Take your learnings about what worked and what did not, document them, and apply them to the next campaign.
Second, make sure that you have processes for your overall marketing mix, across all campaigns and channels. This approach will let you see patterns that don’t depend on the particulars of one particular product, promotion, time of year, or other critical but external factors. The best approach here is to do quarterly reviews and annual reviews of all multichannel marketing efforts. These periodic reviews can feed into your future marketing plans and budgets.
For both in-flight and periodic reviews, there are two critical factors for success.
One, develop in advance a list of specific questions that you’ll use your attribution data to answer. These questions help you with the planning. They also help you compare campaigns or time periods to each other so that you continually improve your results.
Two, whether you are a marketing team or a one-person shop, challenge your assumptions aggressively. If one channel is performing relatively better than others, ask yourself tough questions about what factors might have contributed to that performance and about what else might have caused it. Maybe your social media ads overperformed because you had a better-targeted audience, because you changed your call to action, or because your promotion is a better offer than your competitors. The key thing isn’t to jump to any conclusions.
A Quick Guide to Platforms
If you search on Google for marketing attribution tools (our advice—don’t do it!), you’ll find what seems like countless options for tools and platforms that allow you to do attribution modeling, click attribution, touch attribution, and many other similar options.
Chiefmartec.com, a blog that focuses on marketing technology, has identified 8,000 different marketing tech platforms. More than 200 of these are positioned as attribution tools. The right tool for your e-commerce business depends on the specific channels that you use for marketing, the readiness of your data, the skillsets of your marketing team, the complexity of your plans for using attribution modeling, and, last but not least, your budget.
There’s no one-size-fits-all, best-for-everyone recommendation that we could make for you, and if you do search for them, take any recommendations with a grain of salt. In fact, many of the reviews you’ll find are posted by the actual vendors of an attribution tool.
To evaluate platforms, ask yourself a few important questions.
- Does it integrate with your customer database, your email platform, your PPC, social, and display ad platforms, your SEO tool, and your web analytics? Make sure that integration is listed as an out-of-the-box feature and not just as a potential customization.
- Is it secure and does it conform to data privacy standards? Because your attribution tools handle personal information, you should take security and privacy very seriously.
- Is the attribution data reliable? Attribution can be complex, and in addition, some advertising platforms can give inaccurate, incomplete data. In some cases, your platform will even need to filter out click fraud. According to ClickCease, 20% of clicks on PPC ads are fraudulent, and 50% of ad impressions on Internet Explorer were non-human traffic.
- Are the tools easy to use? It doesn’t necessarily make sense to purchase a high-power data analytics platform with every bell and whistle, every possible dashboard view, and more. Instead, purchase what makes sense for your business goals and for the skills of your marketing team.
- Is the support model appropriate for your needs? Some companies offer standard and additional support that help you get started and make the most of your attribution platform. If you need to onboard yourself and have never worked with similar technologies in the past, you should be sure that your vendor offers what you need during and after your onboarding.
- Is the tool meant for e-commerce? Not every tool is ideal for every industry. Some have a more B2B slant, some are better for services rather than products, and some are more suited for manufacturers of consumer goods than for retailers. Your vendor doesn’t need to specialize in e-commerce but should be able to point to meaningful experience with it.
- Is it a viable company? This area of technology sees a lot of innovation from clever startups. However, you don’t want to risk investing time, effort, training, data, and money in a platform that may not be around in a year or two. You don’t necessarily have to go with an enormous company such as Google or Salesforce to mitigate this risk, but a vendor’s health and longevity should be a part of your due diligence checklist.
In doing your due diligence, there are also four main types of vendors that provide attribution analytics.
“Free-standing” attribution platforms are separate tools that you install and set up within your overall infrastructure, in addition to the tools you already have in place for marketing, CRM, and web traffic analytics. In many cases, these platforms are offered by large and well-known companies. They are designed to bring in data from many sources, but as a result, they can be complex and difficult to understand at first. Examples include Funnel, Impact (formerly known as Altitude), Neustar, Nielsen Marketing Attribution (formerly known as VisualIQ), and Ruler.
Many marketing automation tools also include attribution features. The advantage of this integration is that you get a one-stop-shop for creating and measuring marketing assets, as well as strong CRM functionality. In some cases, however, these tools have less ability to measure marketing assets that aren’t created from within their platform (such as digital display ads or social media) and less coverage of customers who don’t already have records in the CRM. Examples include ActiveCampaign, EngageBay, HubSpot, Infusionsoft, Marketo, and Salesforce.
Tracking and analytics tools also can be used to perform attribution modeling and analytics. The best known of these, of course, is Google. The free version of Google Analytics has some basic attribution modeling built-in. Google offers a paid platform called Analytics 360 that has much more extensive attribution functionality. Google Analytics also offers a dedicated enhanced e-commerce plugin. While some tech industry experts point out that Google may not give you a comprehensive view of social media interactions (such as Facebook’s robust ad platform), it’s possible to do so more effectively by using the right tracking codes embedded in your links). Other alternatives in this category include Chartbeat, StatCounter, and Woopra.
Finally, the top e-commerce platforms also offer attribution analytics, including BigCommerce, Shopify, and WooCommerce. They also have additional analytics and attribution tools that you can find in their Integrations or Marketplaces (i.e., approved and vetted tools that work well with your e-commerce platform.
The Art and Science of Doing Your Best
Attribution can become a valuable part of your overall analysis of marketing effectiveness, but it also can be very complicated to get right. It requires a lot of investment above and beyond the cost of the tool that you use to do it.
You have to change the way you execute marketing campaigns in order to send the right data to your attribution models. You have to learn how to use tools with a lot of complex features. You also need to develop data-centric processes and take a nuanced approach to the ways that you understand and act on the data and reports that your attribution tools generate.
Most importantly, you have to keep in mind that marketing attribution is as much an art as a science, despite all the sophisticated technology, data modeling, and statistics. It takes time and practice to figure out the best way to bring attribution into your organization — careful assessment of what you know and do today, thoughtful review of your marketing channels and existing technology, and realistic expectations about what you can and can’t expect.
You won’t get your attribution model or reports right the first time. You have to live with the potential blindspots around offline interactions, anonymous browsing, lack of data about new customers, and more. But if you don’t get distracted by chasing perfection, and if you make careful use of the data to influence rather than dictate your marketing decisions, you’ll find that attribution helps you get better results by doing more intelligent marketing over time.