How to turn attribution data into actionable insights
The power of actionable measurement
In 2019, Google teamed up with Boston Consulting Group, a leading management consulting firm, to study the digital marketing practices of 200 companies around the globe. Their goal was to identify the traits and technologies that defined today’s most advanced and successful marketers, and to measure the value of these best-in-class characteristics.
Among the most successful and digitally mature organizations, several common characteristics emerged, including a concept that should be front-of-mind for digital marketers everywhere: each of the companies demonstrated “actionable measurement.”
According to the study, actionable measurement occurs when: Marketing objectives are linked across channels to unified brand goals; Objectives are validated by sophisticated digital marketing attribution; and Attribution data is used to drive smart, targeted action.
The benefits of marketing systems based on actionable measurement were clear: companies that the study designated as successful in this regard drove an average of 20% greater revenue at 30% lower costs through digital marketing than their less advanced peers. One company evaluated in the report proved the point even more plainly:
After collecting insights from data-driven attribution and using them to take quick and appropriate action by optimizing ad budgets and keyword bidding, the company saw an immediate increase in lead volume of 6% and a reduction of cost per lead of 17%.
Luckily, collecting attribution data has never been easier. Unfortunately, collecting data and knowing what to do with it are two very different animals, and attribution data is only valuable when it drives targeted action.
In the sections that follow, we’ll take a look at how to achieve a marketing system based on actionable measurement powered by sophisticated attribution. Starting with the definition of actionable measurement included above, we’ll examine:
- How to create marketing objectives that are linked across channels
- How these attribution data to measure these objectives is collected and analyzed
- How attribution data can be used to drive action and optimize systems
Linking marketing objectives to unified brand goals
Step one of implementing a marketing system based on actionable measurement is to create a framework in which your data can be put to use. To do so, your organization should:
- Make a plan based on the overall brand goals of your business
- Create marketing objectives that support that plan
- Evaluate those objectives by measuring key performance indicators (KPIs) supported by attribution data
Brand goals are the big-picture priorities for your business. They include the highest-level plans of your business model and strategy. These may involve rolling-out new products, expanding into new markets, or scaling up operations and production. They are generally tied to financial measures such as growth targets and profit margins.
Marketing objectives are your plans for utilizing marketing efforts to achieve your brand goals. They describe what your marketing team intends to do and how, including clear direction for team members on implementing plans, and provides reporting you can share with ownership or executive teams to support oversight. Examples of marketing objectives include growing your online presence, generating new leads, increasing sales, and building brand awareness, among many others.
Key performance indicators (KPIs) are quantifiable metrics that measure the performance of your marketing efforts. Depending on your marketing efforts, your KPIs may include sales growth, changes in profit, customer lifetime value, conversion rates, website metrics, and SEO performance.
A common set of guidelines for creating marketing objectives is called SMART. It’s an acronym that incorporates the characteristics of successful marketing objectives. Objectives should be Specific, Measureable, Achievable, Relevant (related to brand goals) and Time-bound.
Validating marketing objectives with attribution data
Your marketing objectives should be designed to support your brand goals, and your marketing activities should be aligned to support your marketing objectives. Furthermore, these activities should be aligned across marketing channels—the best marketers build an in-depth understanding of the entire customer journey, and they know how and where to coordinate and focus their engagement effort.
This gets us to our second step: validation of objectives via attribution data. Your attribution system should be designed to provide data that supports your ability to evaluate the progress you’ve made on your marketing objectives across channels. However, raw attribution data—clicks, ad impressions, website traffic, email opens, bounce rates, etc.—might not provide a full enough picture to serve as useful KPIs on their own.
How much is each click worth in terms of impacting your bottom line? How many users who fill out an online form eventually become customers? How often does an opened email lead to a purchase?
In order to answer these questions, let’s take a look at some marketing attribution basics.
What is marketing attribution?
Marketing attribution is the process of determining how much credit each of your marketing efforts deserves for influencing a consumer’s decision to make a purchase (or take some other action that drives value for you business). More specifically, attribution involves collecting data about the activities of your customers whenever they interact with media related to your business, and based on this data, assigning value to various marketing channels and touchpoints based on how much influence they had in causing a customer conversion.
The attribution process is based on the idea of the customer journey. The customer journey follows the consumer’s path from awareness to intent to conversion. It begins when a person develops a desire for a product or service and becomes aware that your business offers what they’re looking for; it gets longer as they interact with various materials designed to encourage the action that you want them to take; and eventually it ends when the person becomes a customer. If the customer's lifetime value (LTV) is considered, however, and there are additional purchases after the intial purchase, then the journey continues on.
Touchpoints are the ways in which people interact with your brand, including all the various media materials, experiences, and activities that make up your business’s marketing efforts. A touchpoint can range from a banner ad on a website, to a billboard on the side of the highway, to a conversation with a sales representative in a store or over the phone.
Channels are the various types of media through which you deliver your message. Think of them as being the vessels that bring your touchpoints to your target audience. Social media platforms, various types of advertising (such as Google search ads), television advertising, email marketing—these are all types of channel, among many others.
Conversions are any actions that a customer takes that drives value for your business by contributing to that customer’s lifetime value (LTV). Just like it sounds, lifetime value is the monetary amount that a customer contributes to your bottom line over their lifetime as a customer—from their first purchase to their last.
While the most common way to contribute to LTV is by making a purchase, a value-adding conversion may take the form of other types of desired action as well; for a social media platform like Facebook, accounts are free and revenue is driven by advertising. Therefore, signing up for an account might count as a conversion, even though no direct revenue is generated at that point. The customer can now be targeted with ads, so this conversion still contributes to that customer’s LTV.
How to estimate lifetime value
Sometimes a customer’s LTV is defined by a single purchase: a shop that sells memorabilia to tourists, for instance, may sell a t-shirt to someone visiting on vacation who will never return to the area, much less the shop. If the t-shirt costs $10 and the profit margin in 75%, that customer’s LTV is $10 x .75 = $7.50.
On the other end of the spectrum, the customers at a dentist’s office would have LTVs that comprise multiple conversions: in addition to regular, bi-yearly cleanings, customers come back and spend money for every cavity and chipped tooth, producing a multi-transaction LTV in the tens of thousands of dollars (minus expenses) spanning many years.
Imagine you run a mail-order auto parts business called Parts and Parcels that buys auto parts wholesale from OEMs, sells them on a website, and sends them to consumers via the postal service.
Based on calculations by your accounting department, you know that the average value of a sale is $700, at an average profit margin of 50%. Armed with this data you can calculate your short-term conversion value: $700 x .5 = $350.
This $350 figure is a nice estimate that can get you started in taking action in optimizing your budget allocation. At the same time, there are other factors to consider in calculating conversion value in addition to the one-time monetary reward of making a sale. Three common concepts to keep in mind are:
- Repeat business (aka LTV)
- The word-of-mouth impact of gaining a new customer
- The cost of maintaining a customer relationship over time
Consider again your mail-order auto parts business. In addition to the $700 value of the initial sale, the accounting department has determined that the average customer spends an additional $2,300 during the lifetime of their relationship with the business. The LTV of the average customer would therefore be $700 + $2,300 = $3,000 x .5 (profit margin) = $1,500.
In addition to factoring in total LTV, you may want to consider the word-of-mouth impact of converting a new customer. In many cases, new customers can be the best evangelists for a business that serves their needs and treats them well—creating a significant impact in recruiting new business. So, assuming Parts and Parcels has calculated added business due to word of mouth as 15%, you can calculate the following: $700 + $2,300 = $3,000 x .5 (profit margin) = $1,500. Add in a 15% added value by multiplying by 1.15 and you get a total conversion value of $1,725.
Finally, you know that the average customer requires upkeep in the way of occasional customer service interactions. You crunch the numbers on the hourly rate of your customer service representatives and find an average lifetime maintenance cost of $225. Subtract that from your total conversion value of $1,725 and your new conversion value is $1,500.
How to collect attribution data
Digital marketing as a whole has exploded in recent years—in 2019, worldwide digital ad spending rose by 17.6% to $333.25 billion, according to eMarketer—followed closely by a significant proliferation of attribution technology. Given the near ubiquity of digital marketing in the media mix of every company, an enormous variety of tools have sprouted up to measure marketing performance based on attribution data.
This technology is based on the ability to track the activity of consumers to illuminate the trajectory of the purchasing journey. It relies on digital tracking tools, such as cookies and tracking pixels, to follow the activity of individuals along the customer journey. The data is collected and analyzed according to various attribution models that attempt to determine how much credit each touchpoint deserves based on its impact in causing the conversion. Based on this type of analysis, marketers can take action to optimize their ad spend, marketing channels, and tactics.
To produce the fullest possible picture of each consumer’s purchasing journey, you need to collect as much data as you can—the more the better. Your view of every buying path should be as complete as possible, as each missing puzzle piece detracts from the accuracy of your analysis when it comes to assigning relative values to each touchpoint.
Attribution data can be gathered by a wide variety of software tools. In general, the most important tools include a web analytics platform, a CRM, integrated advertising technology to automate media buying, bidding, and message development, and tools to capture offline leads such as phone calls.
The most popular web analytics tools come from our friends at Google. Starting with their Google Analytics platform, which launched in 2005, Google began to allow people and businesses to easily track traffic on their websites. This early breakthrough was followed by Google Ads, Google Display Network, and Campaign Manager, tools that provide attribution data about search and display ads, and allow you to see which Google-related touchpoints are responsible for driving consumers to various elements of your online presence. Today, all of Google’s tools are coalescing into a unified dashboard called Google Marketing Platform.
Google has a lot of reach, but it lacks insight into many touchpoints that exist outside of the Google ecosystem. As such, a digitally mature attribution system needs to collect data on many platforms to capture many channels. Attribution tools began as outgrowths of other systems and followed closely with the evolution of digital advertising: most social media platforms, like Facebook, now have sophisticated in-house attribution tracking tools for business users. Other tools are funnel based: CRMs, such as Salesforce, are particularly good at tracking conversions and certain offline touchpoints, such as customer conversations with your sales team. Speaking of offline transactions, tools like CallRail track your interactions with customers over the phone, via calls and texts. Tools like Marketo and Hubspot are all-in-one marketing automation tools that make adjustments based on your attribution data for you.
These are just a few of the options available. Finding the right lineup of software tools is an important first step in creating an attribution system that works for you, but ultimately, the real challenge comes in getting them to talk to one another. You want to create as full a picture as possible of the customer journey, a process that includes integrating data points from multiple tools into an aggregated whole and using it to derive insights across multiple channels at the same time. There are a few tools out there that attempt multi-channel analysis these days, but most companies use custom solutions for combining data from disparate sources.
How to analyze attribution data
The goal of attribution models
The attribution system of every organization exists on a spectrum of digital matury. On one end are the businesses that collect minimal attribution data about the customer journey and make more-or-less blind decisions in funding and developing their marketing strategies. On the other end of the spectrum are businesses with fully integrated, multi-moment marketing systems that drive single-customer business outcomes. These systems are digitally advanced and succeed by delivering the right content to the right consumer at the right time.
No matter where your organization rates on the digital maturity scale, your attribution process can always be improved. The goal of any data-driven marketer is to construct a model that most accurately reflects reality, and then, based on the information generated by that model, efficiently make adjustments in overall ad spend, spend distribution across marketing channels, and intra-channel marketing tactics. Ideally, every single contributing factor in a customer’s decision to make a purchase is accounted for and accurately weighted according to its impact on driving value for the business, and the appropriate adjustments are made instantaneously.
Don’t expect to achieve a perfect model of reality in practice—the full complexity of the human decision-making process includes far too many variables that are invisible or immeasurable—but the pursuit of the ideal is an important conceptual starting point in thinking about how to build attribution-driven systems that work for your business. No matter where you land on the digital maturity spectrum, any attempt at a more accurate understanding of the purchasing journeys of your customers moves you in the right direction and has positive ramifications for your business.
What are attribution models?
Once you’ve collected raw attribution data, it’s time to begin your analysis. As discussed above, the processing of attribution data involves assigning credit to various touchpoints; determining how much credit each touchpoint should be assigned makes up the bulk of attribution analysis.
There are two primary approaches to attribution: touch-based attribution and Media Mix Modeling.
Touch-based attribution models track each consumer along their individual customer journeys and assign credit to each touchpoint based on where on the purchasing path it’s situated relative to the others.
Single-touch attribution models choose one specific touchpoint on every path—the last click before a purchase is made, for instance—and give it all the credit for causing a conversion. Multi-touch attribution models distribute various levels of credit across all the touchpoints based on their position in the customer journey.
Touch-based attribution models are easy to understand and apply without sophisticated technology or mathematical models. There are a ton of useful primers on touch-based attribution on the web that can get you started. Applying these models to your attribution data is a great first-step to achieving actionable measurement. They provide starting-point data to give you a sense of where to begin.
For digitally mature attribution systems, however, touch-based attribution models with set formulas for assigning credit based solely on where a touchpoint is positioned in the customer journey leave a lot to be desired. The flaws in these models have also been extensively detailed on the web.
To avoid the pitfalls inherent to this type of modeling, the most successful digital marketers have become more sophisticated. By applying machine learning to your touchpoint analysis based on consumer behaviors, you can achieve a much more accurate type of attribution modeling—without the blind spots of standard touch-based models.
What is a data-driven attribution model?
Data-driven attribution (DDA) modeling takes the guesswork out of attribution models by using actual consumer behavior as the basis for assigning credit to various touchpoints. DDA collects all of your tracking and conversion data on every customer journey for which you’ve got records. Using predictive algorithms based on machine learning, DDA determines a more accurate accounting of which touchpoints actually lead to conversions, and makes constant updates based on new input.
While DDA is a more accurate and cost-effective way to model attribution data—studies have shown that marketing programs optimized by DDA models generate more conversions with the same budget levels as those with rule-based attribution models—there are challenges to get it up and running.
For one thing, you need a wealth of conversion data in order to process a large enough sample size for DDA to work its magic. It doesn’t start working out of the box, and generally needs a good amount of lead time to produce reliable results. Many businesses that use DDA analysis started out with various iterations of multi-touch, rules-based models until enough data had been collected to implement DDA.
Another issue involves integrating data from a variety of sources. While Google Ads offers the ability for users to process data using a DDA model, any touchpoints that occur from sources outside the Google ecosystem—like Facebook ads, for instance—are not accounted for in the model. Going back to the scale of digital maturity, the most mature companies find ways to integrate all of their data into a unified system, often by investing significantly in in-house programs and custom solutions.
What is Marketing Mix Modeling (MMM)?
Instead of tracking the individual customer journey, MMM uses aggregate sales and marketing data to assign credit to marketing efforts for driving conversion. It involves collecting historical data based on sales over time and using multivariate regression analysis—a type of sophisticated statistical analysis—to determine how much impact marketing had on customer behavior.
As a top-down model, MMM needs a huge amount of data and complex algorithmic calculations to arrive at its conclusions. It relies on historical data regarding seasonality, pricing, and broad economic conditions.
While MMM can be useful and accurate, it is cumbersome and expensive. Typically, enterprise vendors with footings outside of marketing. It is also not as flexible and responsive as touch-based modeling: companies typically run MMM analysis on a quarterly or yearly basis.
How to use attribution data to take action
At the highest level, an effective attribution system can support decision-makers in determining how much overall budget to allocate to marketing activities. These decisions, which typically occur on a yearly or quarterly basis, are driven by top-level business considerations, such as profit margin targets and growth projections, and are significant factors in the overall fiscal health of a business.
The foundational bit of data in making these decisions is cost per acquisition. Cost per acquisition is the monetary amount you spend on marketing efforts for every new customer those marketing efforts convert. To arrive at a starting point for this number, use your attribution data to total all of your conversions, then take how much you spent on marketing over a given period and divide it by how many conversions you gained in that period.
However, for a more accurate accounting of cost per acquisition, an important concept to keep in mind is incrementality. Incrementality measures conversions driven by marketing in the context of all new conversions.
Incrementality accounts for conversions that would have occurred naturally without marketing. The incremental value of your marketing is calculated by subtracting the business you gained from marketing efforts from the total amount of new business. This way, you can accurately calculate the true impact of your marketing efforts in driving business outcomes.
Marketing Mix Modeling, as described in the section above, is one way to arrive at an accurate, incremental cost-per-acquisition. Another is to employ experimental methods such as switchback and synthetic control.
In the switchback method, a marketer shuts off all marketing for a period of time and compares the amount of conversions during that time with the amount of conversions from a time when marketing was on. In synthetic control, marketing is shut off for a certain market—such as a specific geographical area—while remaining in place for a different, similar market, and conversions are compared.
These types of experiments help you separate conversions driven by marketing from those that would have occurred naturally, improving the accuracy of your cost-per-acquisition calculation.
Armed with LTV and cost-per-acquisition data, you can make solid macro-budget decisions. If your goal is to invest in marketing to grow your revenue by a certain amount, you can use LTV to determine how many customers you would need to gain to hit that revenue, and then multiply your cost-per-acquisition by that amount to reach your growth goals. And by including LTV, you know the highest level of cost-per-acquisition spend that is allowable to maintain profitability. If your cost-per-acquisition goes over that number, your revenue may grow, but you’ll end up losing money.
Optimizing channels and touchpoints
The next levels in which attribution data can direct action is on a per-channel and per-touchpoint basis. Here’s where multi-touch and data-driven attribution models come into play. By evaluating the performance of channels and touchpoints and attributing values to them based on their efficacy in driving conversions, you are armed with the data you need to make smart, channel-based allocation decisions and optimize the timing and content of your touchpoints.
The insights generated through attribution ultimately help drive the incremental value of each marketing effort. When you properly assign impact, you can distribute funding based on the mixture of touchpoints that most efficiently work together to drive conversions, thereby lowering the marketing cost your business incurs for each conversion.
Channel-based insights are typically based around allocation decisions. How much should you invest in each channel? We covered the analysis behind these decisions for the most part in the sections about touch-based modeling: understanding the relationships between channels depends largely on how you weight your touchpoints, and where those touchpoints live.
Say you’re running a campaign across multiple channels, and one of the customer journeys you’re tracking goes like this: they open an email, then visit your website, later on, they click on an ad before a youtube video, then give you a call based on the number on your website and finally make a purchase. Each touchpoint exists in a different channel, and your allocation decision depends on how much weight that touchpoint was assigned by your model. The benefit of a data-driven model is that it tells you, based on an aggregation of customer decisions, which touchpoints appear to be most effective, taking out the guesswork.
Touchpoint-based insights tend to involve optimizing creative decisions and content. By categorizing the content of your touchpoints based on topic and funnel depth, and using the content grouping feature of Google Analytics, for instance, you can gauge the performance of various touchpoints among various demographics of site visitor.
CTAs (Call to Action) are another example. These are short, bold terms, placed on a link or button, designed to initiate action. By creating a conversion event in Google Analytics based on people clicking CTAs, and then comparing conversion rates of different CTAs against one another, you can get a good idea of which are most effective.
Ultimately, the goal of actionable measurement is to create a marketing system that delivers the right message to the right customer at the right time. Building this type of system requires the constant collection of attribution data that is then fed back into the system to optimize every channel and touchpoint. By making sure marketing objectives are linked across channels to unified brand goals, validating objectives with sophisticated modeling, and using attribution data to drive action, you can become a digitally mature marketer.