Download pdf: Algorithmic Attribution SESChicago2013
As published by Adotas 6/22/12 (view article)
Digital media attribution is a hot topic, but it’s still a confusing concept for many. Given a lack of industry standards, and an ever-expanding list of attribution methodologies, it can be difficult for marketers to determine exactly what they need. This post intends to educate and enable marketers to optimize digital spend through advanced insights.
The Funnel and Attribution
Simply defined, attribution is allocating credit to each interaction that drives a desired action (visit, goal page view, conversion, etc.). Within this broad definition, there are two primary distinctions:
• Lower-funnel or click-based attribution incorporates assist clicks when allocating credit for conversions. Compared to last-click reporting, this is a step in the right direction. The limitation to lower-funnel attribution is that it severely discounts the role of display advertising while overstating the role of search, affiliate, email and other click-centric media. For online advertisers seeking a more complete picture, a full-funnel view is required.
• Full-funnel attribution builds on click-based attribution by incorporating assist impressions from display ads (video, rich media, flash and .gifs) when allocating credit for visits or conversions. Recognizing that display ads can be very effective, even in the absence of clicks, a full-Funnel attribution model is needed to quantify the true impact display ads have in creating awareness, consideration and preference.
Cross-channel attribution addresses the role of each digital channel (display, paid search, natural search, email, affiliate, etc.) plays in the customer engagement process. While conversion paths are interesting, they aren’t very actionable. To truly understand and optimize each channel, you must allocate fractional credit to each channel and placement that contribute to a measurable action. This generally results in a shifting of credit from non-paid channels (organic search, direct navigation, referring sites), back to paid media (display, paid search, email, etc.) that “fed” the non-paid channels. Before diving into weighting methodologies, let’s first look at leading approaches to attribution.
Three Approaches to Attribution
While there are many approaches to attribution, here are three you should be familiar with:
• Statistical attribution is based on traditional media-mix modeling, which relates to analysis of disparate data sets. In this approach, you would analyze three months of impression data and three months of conversion data and look for relationships between the data sets. At best, this approach can provide high-level directional signals. If you want granular insights into the impact of each channel, vendor, placement or keyword, you need a more granular approach.
• A/B testing seeks to attribute credit and validate causation by observing results from pre-defined combinations of media placements. A/B testing can be used to measure display’s impact on results from search, as well as the performance of a specific creative, vendor, market or channel. While A/B testing is a great way to observe directional insights, it’s nearly impossible to exclude or account for other factors (seasonality, competitors, macro-economics, weather, etc.) that might impact performance between the control and test groups.
• Operational attribution takes a bottom-up (visitor-based) approach to analyzing and allocating credit to impressions, clicks and visits that precede each conversion. With operational attribution, there is no need to calculate possible conversion paths – you have the actual data, which provides for a more granular and accurate data set for advanced analysis. (i.e. heuristic and/or statistical modeling). As with any approach, caution must be exercised to define the appropriate look-back window and cleanse the data set and exclude factors (e.g. wasted impressions and post-conversion visits) that might skew the results.
Manual vs. Statistical Weightings
With the framework of operational attribution, weighing assist impressions and clicks is both art and science. Here are two primary approaches:
• Subjective weighting of impressions and clicks: First-generation platforms require the marketer to define the rules for allocating credit to each interaction. While this approach is easy and flexible, it lacks statistical rigor and entails too much guesswork. Moreover, it allows the operator to influence outcomes through the assumptions that drive the model. But the biggest problem is that it allocates credit based on the actual number of assist impressions, rather than using observable data to model how many are actually needed. For example, you may find that 12 impressions preceded an average conversion, when only six impressions were actually needed. If you give credit for wasted impressions, you end up rewarding vendors for over-serving customers.
To reduce subjectivity and improve accuracy in how impressions and clicks are weighted, we must use machine learning and algorithmic modeling.
• Algorithmic weighting: To remove the guesswork in attribution can use machine learning and proven algorithms to calculate probability-based weightings for assist impressions and clicks. This removes the arbitrary nature of manual weightings and provides much higher levels of confidence and comfort for marketers.
There are numerous approaches to statistical modeling and there are plenty of vendors vying for the “best math” award. While it’s hard to say which approach is best, we have seen that most prefer transparency vs. opacity, and known algorithms to proprietary models. If you’re going to put your neck on the line to defend a new measurement standard, you should be comfortable with the approach and underlying assumptions. In general, a transparent, statistically validated approach is best.
There is an endless number of attributes you can seek to measure, including segment, format, audience demographics, creative concept, message, frequency, day-part, sequence, etc. While the idea of Metric Nirvana is appealing, it’s also very elusive if you try to get there overnight.
It’s important to recognize that the more granular you get, the more data you need and the more complex the setup, production, reporting and analysis. Many attempts to go directly from last-click to advanced micro-attribution fail due to the complexity of implementation and analysis. We all crawled before we walked, and walked before we ran. Analytics should be no different.
If you’re still using last-click metrics, start with channel-level and publisher-level attribution. Once you’ve identified your top performing vendors, delve deeper sequentially (as opposed to all at once) by looking at recency, format, creative and other variables that may have incremental impact on the end results. Just remember that with each incremental value, the scale, complexity, and risk of failure increase dramatically. So start with the low-hanging fruit, and work your way up the tree.
To sum all this up, we are Encore and MediaMind advocate the following best practices:
• Operational, full-funnel and cross-channel attribution.
• Use machine learning to weigh and allocate fractional credit to assist impressions and clicks.
• Improve your chances of success by starting with basics and getting more granular over time.
Hopefully you now have a better understanding of the attribution landscape and some of the distinctions within it. As always, feel free to comment, tweet, like, post, share, or whatever it is you do in your own social sphere. Thanks for stopping by!
Last month I had the pleasure of moderating the Display Ecosystem panel (View the Video) at Rapleaf’s 2012 Personalization Summit. On my panel were experts from leading companies that represented numerous categories within the display landscape. Panelists included:
Our discussion addressed many of the issues that we are grappling with in the Ad-Tech industry, including:
After our discussion, I thought about the implications for the Display Ad ecosystem, and for the Ad-Tech industry as a whole. Here are a few of my thoughts…
The wall mural below sums up the discussion – and made for a nice graphic snack for attendees.
As always, feel free to comment, tweet, like, post, share, or whatever it is you do in your own social sphere. Thanks for stopping by!
In April 2012 I presented a case study on Full-Funnel Attribution at the granddaddy of all industry conferences: Ad-Tech in San Francisco.
I was honored to share the stage with Young-Bean Song, a pre-eminent thought leaders in digital media measurement and analytics (and a very nice guy). After years of applying to speak at Ad-Tech, I was finally selected; not because I’m the world’s most pre-eminent speaker but because the case study we developed is so effective at presenting how advanced analytics and full-funnel, cross-channel Attribution can be utilized to maximize performance and boost Return On Spend.
Among the highlights of the case study, we demonstrated:
For those who didn’t make the show, I’m happy to share the case study in two formats (both are hosted on slideshare):
If you’d like to learn more about Attribution or discuss the case study, please drop me a line (see Contact link below). Also please feel free to comment, tweet, like, post, share, etc. as you see fit. Thanks for your time and interest!
Today a colleague sent me a link to a new article on Attribution and media measurement with a request to share my thoughts. Written by a statistician, it was the latest in a series of published perspectives on how Attribution should be done. When I read it, several things occurred to me (and prompted me to blog about it):
In reading the comments below the article, my mind drifted back to business school (or was it my brief stint in management consulting?) and the theoretical discussions that took place among pontificating strategists. And then it hit me… even in one of the most innovative, entrepreneurial and growth-oriented industries, an Ivory Tower mindset somehow still persists in some corners of agencies, corporations, media shops and solution providers. Not afraid to share my views, I responded to the article in what I hope was a polite and direct way of saying “stop theorizing and focus on the real problem.” Here is my post:
“…We all agree that you need a statistically validated attribution model to assign weightings and re-allocate credit to assist impressions and clicks (is anyone taking the other side of this argument?). And we all agree that online is not the only channel that shapes brand preferences and drive intent to purchase.
I sympathize with Mr. X – it’s not easy (or economically feasible) for most advertisers to understand every brand interaction (offline and online) that influences a sale. The more you learn about this problem, the more you realize how hard it is to solve. So I agree with Mr. Y’s comment that we should focus on what we can measure, and use statistical analyses (coupled with common sense) to reach the best conclusions we can. And we need to do it efficiently and cost-effectively.
While we’d all love to have a 99.9% answer to every question re: attribution and causation, there will always be some margin of error and/or room for disagreement. There are many practitioners (solution providers and in-house data science teams) that have studied the problem and developed statistical approaches to attributing credit in a way that is more than sufficient for most marketers. Our problem is not that the perfect solution doesn’t exist. It’s that most marketers are still hesitant to change the way they measure media (even when they know better).
The roadblocks to industry adoption are not the lack of smart solutions or questionable efficacy, but rather the cost and level of effort required to deploy and manage a solution. The challenge is exacerbated by a widespread lack of resources within the organizations that have to implement and manage them: the agencies who are being paid less to do more every year. Until we address these issues and make it easy for agencies and brands to realize meaningful insights, we’ll continue to struggle in our battle against inertia. For more on this, see “Ph.D Targeting & First Grade Metrics…”
I then emailed one of the smartest guys I know (data scientist for a top ad-tech company) with a link to the article and thought his reply was worth sharing:
“I think people are entirely unrealistic, and it seems they say no to progress unless you can offer Nirvana.”
This brings me to the title of this post: It’s hard to solve problems from an Ivory tower. Note that this is not directed at the author of the article, but rather a mindset that persists in every industry. My point is that arm-chair quarterbacks do not solve problems. We need practical solutions that make economic sense. Unless you are blessed with abundant time, energy and resources, you have to strike a balance between “good enough” and the opportunity cost of allocating any more time to the problem. This is not to say shoddy work is acceptable; as stated above, statistical analysis and validation is the best practice we preach and practice. But even so-called “arbitrary” allocation of credit to interactions that precede conversions is better than last-click attribution. It all depends on your budget, resources and the value of advanced insights. Each marketer needs to determine what is good enough, and how to allocate their resources accordingly.
Most of us learned this tradeoff when studying for finals in college: if you can study 3 hours and make a 90, or invest another 3 hours to make a 97 (recognizing that 100 is impossible), which path would you choose? In my book, an A is an A, and with those 3 additional hours you could have prepared for another test, sold your text books or drank beer with your friends. Either way, you would extract more value from your limited time and energy.
To sum up, we need to focus our energies away from theoretical debates on analytics and media measurement, and address the issues that prohibit progress. The absence of a perfect solution is not an excuse to do nothing. And more often than not, the perfect solution is not worth the incremental cost and effort.
As always, feel free to comment, tweet, like, post, share, or whatever it is you do in your own social sphere. Thanks for stopping by!
In early June I was fortunate to be one of 350 ad tech CEOs who attended LUMA Partners’ Digital Media Summit in NYC, featuring the best and brightest in the industry. I’ve been to some great networking events before (IAB, 4A’s, etc.) but this was tough to beat.
In addition to meeting some amazing people, one of the highlights was the release of the latest display ad landscape or “LUMAscape” aka “the slide” that was originally produced by Terence Kawaja in 2010. For those who are new to display advertising (or have been out of the market for the last 3 years), buying display media is like buying a house: you also need phone service, internet, cable, gas, electricity, dog-walking, etc. In this case, Media is the house; ancillary services include ad verification, OBA compliance, data/tag management, audience measurement, ad serving, and our favorite: attribution.
The newest version of the slide is getting ever closer to accurately depicting all the segments and sub-segments that comprise the digital advertising landscape. It also marked the debut of Encore Media Metrics as a recognized leader in the Attribution and Measurement category.
The industry is extremely fragmented, and is likely to stay that way for a while. So if you want to play in the display advertising space (either as a buyer, seller or manager) you need to understand the difference between a DSP, DMP and SSP without yelling “WTF!” Yes, it’s easier said than done but this map should help you get started.
Steve Latham (@stevelatham)
Advertisers are (finally) looking beyond the last click. Here is an overview of the Five Forces that are driving adoption (also published by MediaPost in May 2011)
It’s been 3 years since measurement buzzwords “attribution” and “engagement mapping” emerged with great anticipation and excitement in online advertising. The idea of looking across digital channels and beyond the last click to measure media throughout the funnel was thought to be the holy grail in online marketing. Recognizing that “last-click wins” is insufficient for measuring the brand-building attributes of display media, brands, agencies and media vendors saw Attribution as the next big thing in digital advertising.
Yet as we entered 2011, very few marketers were using Attribution to measure and optimize online media spend. Despite the universal desire for better measurement, most were still using old metrics (click-through rates, cost per click and direct cost per action) to analyze paid media. Greg Papaleoni, who develops Analytics and Insights for Yahoo! Advertising Solutions, sums it up well: “While Full Funnel Attribution is the future of the ever-evolving digital media measurement landscape – it should be the present. Those advertisers who embrace and implement this logic, methodology and technology sooner rather than later will enjoy a massive advantage over their competition.”
While adoption has been slow to date, this is changing quickly due to the convergence of numerous factors. Borrowing on Michael Porter’s “Five Forces” model for analyzing industries, here is my take on the Five Forces that are driving digital media attribution (author note – I received permission from Professor Porter to adopt his model to this category):
1. The continuing shift of media budgets from traditional to digital.
While total U.S. media spend will grow only 3% in 2011, digital spend will grow 14%, surpassing Newspaper as the #2 medium. Accounting for almost 30% of daily media consumption, Digital spend will continue to outpace all other channels for the foreseeable future.
2. The resurgence of display advertising
Per eMarketer, Display media spend will grow 14% in 2011, outpacing 10.5% growth in paid search. While there are many reasons behind the growth (consumption of social, video and mobile content, better targeting capabilities, real-time bidding, richer formats, etc.) I believe the resurgence of display is driven by two primary factors:
3. Increasing focus on accountability
While marketing budgets may have loosened, the focus on results has not. As a result, marketers are keeping a very close eye on ROI from “brand-building” media. With the ever-increasing need to show ROI, brands now want branding plus performance. To properly measure brand-building media, we need to measure engagement, not clicks.
4. Evolution of web architecture
Recent forays by IBM and Oracle into the marketing arena signal a new wave in convergence of IT and Marketing. As the IT behemoths push technology-based marketing solutions, CIOs are becoming more attentive to the needs of the marketing department. The deployment of Data Management and Universal Tagging Platforms enable advanced analytics and media measurement that were off-limits to marketers in the past. With this roadblock removed, the stage is set for new measurement tools to be deployed across their digital infrastructure.
5. The emergence of better Attribution solutions.
While early Attribution solutions were expensive and limited in capabilities (e.g. couldn’t attribute credit for organic conversions), a new breed of point-solution vendors (including my company Encore Media Metrics), are now offering more effective, flexible and affordable solutions. For a very modest investment (as low as 1-2% of media spend), advertisers can now have a much more holistic and accurate view into the performance of each channel, vendor, format, placement and keyword. These insights are enabling advertisers to optimize media budgets, yielding 20-40% gains in revenue. The immediate return on investment in Attribution solutions may exceed 1-20x (100%-2,000%).
The Five Forces Driving Attribution are illustrated below:
As our business objectives change, so must the manner in which we measure results. As dollars continue to flow into digital, brands and their agencies must use more efficient, accurate and effective metrics for measuring media throughout the funnel. The emergence of more advanced and affordable Attribution solutions, supported by growing support from IT departments is paving the way for Attribution to become a foundational component within the digital marketing ecosystem.
Matt Miller, SVP of Strategy & Analytics at Performics, agrees, stating “Attribution is one of the top priorities for us and our advertisers. Focus on attribution will only increase as advertisers build and implement strategies to maximize ROI across all digital channels.”
As always, comments are encouraged. And please feel free to share!
Steve Latham (@stevelatham)
Encore’s CEO was recently interviewed by AdExchanger, a leading online marketing news publisher, about the launch of Encore, the problems we solve and how we’re positioned. You can read the article on AdExchanger or see the transcript below. Enjoy and feel free free to share!
Steve Latham is CEO of Encore Media Metrics, an attribution technology company.
So, what problem are you solving with Encore?
We solve 2 problems for agencies and brands alike. First, we provide advanced attribution and measurement, enabling them to see across channels and beyond the last click to measure performance of paid, owned and earned media. While most marketers are aware of the need for attribution, very few are doing it. Second, we allow them to offload the tedious, manual work of reporting and measurement, which is a loss-leader for most agencies. They need good metrics, but it’s hard to justify the cost of large teams needed to manage all aspects of reporting. We offer a cost-effective way to produce the insights they need to optimize budgets and maximize campaign ROI. So to sum it up, we provide better reports and deeper insights in a way that saves them time and money.
I believe we’re hitting the market at a great time and that this will be the year Attribution goes mainstream for a few reasons. First, paid search is maturing and expanding digital budgets will have to be deployed elsewhere (display, social, mobile, etc). There are only so many searches every day and most companies have optimized their ppc campaigns. The low hanging fruit in search has been picked; further gains will be in much smaller increments and will require buying short-tail terms that start conversations rather close them (hence the need for keyword attribution). This is supported by the fact that Display will grow faster than search in 2011 and is expected to outpace it for coming years. Search is still the big dog, but display and other brand-building media are nipping at its heels. If you believe Eric Schmidt’s prediction of a $200 billion global display market, we’re still very early in this game. Other factors driving Attribution are the increasing focus on accountability, the upgrading of web architecture (e.g. adoption of universal tags) and the emergence of affordable attribution solutions – such as ours. These factors are converging to make 2011 a very exciting year for those of us in the attribution space.
What’s your view on the competitive set and where you’ve been and how would you say you differentiate?
To understand how we’re positioned, you must first understand how the Attribution marketplace is segmented. For starters, there are two different approaches to attribution: operational attribution and statistical or algorithmic modeling. Each approach has its place and I believe they are more complementary than competitive. Statistical modeling analyzes vast amounts of data to look for correlations that indicate how media channels (display, search, email, affiliate, etc.) work together to drive results. Modeling allows you to see which channels feed each other, and which mix should yield the best overall ROI.
In contrast, operational attribution creates detailed records for each visitor that enable you to see which ads were seen and clicked on, how the visitor found your site, what pages they viewed and what actions they took. You can then query the data to analyze engagement paths and assess the performance of each channel, vendor, keyword and placement. We believe operational attribution is the foundation for advanced analytics as it’s based on actual visitor data (vs. a black box) and provides much more granular insights into performance of all types of media. Once you have operational attribution, you can then do advanced modeling of that data to glean additional insights. But, operational attribution will provide 80-90% of the insight you need to optimize your spend.
Within the Operational segment, you then have to look at the extent of attribution: lower-funnel (click-based) vs. full-funnel (clicks and impressions). While click-based attribution is better than nothing, it doesn’t answer the question: “which media buys are creating demand?” The lower-funnel approach relies on clicks, which may be great for search, but insufficient for measuring the impact of display media. If you want a true picture of which ads are creating demand and which placements are satisfying demand, you need a full-funnel solution.
Now to the original question: how are we positioned vs. our competitors? While I can’t speak for our competitors, I can say we differentiate in a few ways: 1) we incorporate attribution from social media (even in the absence of referring clicks), allowing us to provide attribution for paid, owned and earned media, 2) we have a flexible approach that is designed to accommodate varying needs of agencies and brands (no long-term commitments, pay for what you use, etc.), and 3) we are affordable for most marketers. If a client spends between $50,000 and $5 million per month in online media, they can afford our solution.
Is scale of ad spend critical to Encore’s services – attribution, media mix modeling?
If you’re asking is Attribution is only suited for the biggest advertisers, the answer is no. It really doesn’t matter how much you spend; you still need to look across channels and beyond the last click to optimize your mix. Even if you’re only spending $50,000 a month, a small incremental investment can yield a dramatic improvement in Return on Spend. Any advertiser who is buying more than just search is going to benefit from Attribution.
What’s your view on the “view‑through conversion”?
View-throughs are good for ad networks seeking to optimize their media placement, but they are limited in what they offer advertisers. If you are buying display media from 5-6 vendors, you’re likely to get some view-throughs from each buy. While view-throughs tell you if an ad was seen they don’t tell you which ads were the most effective (and cost-effective) in creating demand, or how each media buy influenced results from paid or natural search. You can’t analyze recency or frequency and you can’t tell the order in which ads were viewed. You need more details to truly understand which placements created demand, the role they played in the engagement path, and how to attribute credit within the channel. Yes, you need a full-funnel attribution solution.
What’s the difference between attribution modeling and media‑mix modeling?
In the context of measuring the impact of digital media, they’re effectively the same thing. But for most marketers, media‑mix modeling encompasses all channels, including TV, print, radio and other traditional media. Within that context, operational attribution should play an important role in providing the inputs that go into such a model. We can provide a much more accurate and richer set of data inputs that will enable the global media mix model to produce more relevant and insightful outputs. As mentioned earlier, it shouldn’t be “either / or” when evaluating operational vs. algorithmic attribution. They can work in concert quite well.
What do you see out there as the most difficult channel to provide the sort of service you’re providing today?
Within digital media, Social is definitely the hardest to measure. First, referring clicks are not good indicators as very few actually click-through from social sites to the brand’s web site (see “Connecting the Dots”). But beyond clicks, how do you attribute credit back to people who are watching your You Tube channel, viewing comments on your Facebook page or reading a blog about you? It’s hard because you can’t cookie browsers on 3rd party social media sites. While Facebook now allows marketers to set cookies via iframes on company pages, very few are doing it.
Some try to do social attribution via correlation or looking at directional trends, where a social mentions drove a spike in traffic and a lift in conversions. But this approach is, in technical terms, “squishy.” For most, social attribution is a future goal more than a near term objective.
But since you asked, I should mention that we offer a unique solution to the social media attribution problem. We use a patent-pending tool that allows us to identify which visitors or purchasers have engaged with the brand in social media, regardless of whether or not they clicked through to the site. Through this, we can draw a direct line between online conversions and the social interactions that preceded them. We think it’s pretty cool and we’re seeing a lot of interest from brands, agencies and media vendors.
Do you see social media attribution as an opportunity?
It’s definitely something we see as a differentiator but it should be viewed as part of our solution for two reasons: 1) social should be integrated with other channels from a measurement perspective, and 2) it’s hard to make a ton of money on social media measurement. A brand may spend $500-$1,000 to measure social interaction, but they’re not likely to spend more on the tool than they do on their social media marketing efforts. You also don’t want to be a one-trick pony in the digital landscape. Things move too quickly and one player (e.g. Google) can make render your product obsolete overnight. So we see it as a differentiator and a conversation starter more than a standalone offering.
What is Encore’s target market?
We serve brands and agencies who are seeking to create demand and/or drive sales through paid, owned and earned digital media. While we can accommodate budgets as low as $50k per month, our sweet spot is campaigns with budgets of $100,000 to $2 million per month.
In general, Attribution tends to be more appropriate for considered purchases, e.g. financial, auto, travel, health care, luxury goods and anything B-to-B. The longer the sales cycle and the bigger the ticket, the more you need Attribution.
We work with brands, agencies and trading desks of all sizes, even those with internal ad ops teams. Even if they have a bench, they still need better tools to produce the insights their planners and customers demand.
How does pricing work? Do you charge on according to media spend or is it a per seat?
We price our solution as a technology (vs. a flat % of media spend) that is tiered based on the scope and scale of the campaign. In general, we charge a fixed fee that covers the planning, production and client services, along with a cpm-based fee that covers the cost of data capture, storage and analysis. The fee as a percentage of the media budget will vary significantly. If you’re buying premium placement media at $10cpm, our fees are tiny. If on the other hand you’re going for scale (e.g. $2cpm), the fee will be slightly higher as a percentage of spend. But in either scenario, we’re very affordable and the ROI is hard to beat.
What sort of milestones would you like the company to have accomplished?
My primary goal for 2011 is for Encore to become widely known as a leading provider of measurement, attribution and reporting services. If there is a discussion about Attribution, I want us to be one of the solutions that are always mentioned. Our value proposition (better reports, deeper insights, affordable and adaptable) is hard to beat, and we look forward to proving it to leading brands and agencies.
The What, Why and How of Online Media Attribution
[if you like presentations, view "Attribution 101" on slideshare]
Anyone who has ever bought (or sold) display ads is painfully aware of the need for new metrics for online media. While “last-click wins” may work for paid search, it fails miserably in measuring the impact of display and other media at the top of the funnel. Hence, the need for full-funnel Attribution, which allocates credit for “assists” in the customer engagement cycle.
By attributing credit to contributing impressions and clicks that precede subsequent visits and conversions, marketers can have a much more accurate and holistic view into the performance of each channel and vendor. While most interactive marketers are familiar with Attribution, many are still trying to understand what it is and how it works.
The Need for New Metrics
While digital is the most measurable medium, the “one-size fits all” approach to online media measurement needs to be re-evaluated. While click-through rates (CTRs), cost per click (CPC), direct conversion rates and cost per action (CPA) may be applicable for search and other “bottom-of-the-funnel” media, these metrics are not appropriate or insightful for measuring performance at the top of the funnel, where demand is created.
Display ads can be very effective in achieving their objectives (driving awareness) without any clicks or direct conversions. A recent Media Math study showed that 80% of post-impression conversions are the result of viewing display ads without clicking and only 20% of conversions are the result of a click. In other words, for every conversion that follows a click on a display ad, there are four (4) post-impression conversions without clicks. The upshot: we need better tools and methodologies for measuring the performance of media at the top of the funnel. This is where attribution comes into the picture.
Attribution is the art and science of allocating credit to all interactions that play a supporting role in the customer engagement process. In other words, it’s the act of giving credit for assists. Rather than viewing results from each digital channel in its own silo (a la traditional web analytics platforms), Attribution requires you to take a holistic approach to analyzing how each touch-point contributes to the overall goal (visits, conversions, etc.).
With the resurgence of display advertising, Attribution is becoming increasingly important for optimizing media budgets. As shown in the Google trends chart below show, searches for “online attribution” have increased 150% over the past 36 months.
Approaches to Attribution
Generally speaking, there are two types of Attribution: Operational and Algorithmic / Media Mix Modeling.
We believe operational attribution is the foundation for advanced measurement and analysis of media. The operational approach of giving credit for assists is intuitive, logical and easy to understand. Once the operational attribution model is defined, algorithmic modeling can be used to further optimize the media mix.
Channel Level Attribution
Channel level attribution addresses the relative roles of each media channel in driving traffic and conversions. Attribution requires an algorithm that attributes partial credit to display impressions and clicks that precede visits and conversions. The weighting of impressions relative to clicks will vary based on the type of ad, format, placement and other issues. For example, highly-targeted rich media placements should have higher weighting than Run-of-network animated .gifs. Weightings should be customizable for each vendor and placement.
The channel attribution report below shows the relative impact (last click vs. attributed) of each channel: direct navigation, natural search, referring sites, email, paid search, display advertising and 3rd party email. As shown, attributable credit for display ads may be 50-400% higher than a last-click report would show. It should also be noted that paid search generally sees a net increase in attributable actions as short-tail keywords often play contributing roles in the customer engagement process.
After attributing credit for actions for each channel, spend data can be imported to show the adjusted cost per action for each channel, as shown below. As illustrated, we typically see a 30-80% decrease in attributable cost per action (CPA) for Display, and a slight drop in CPA for paid search (resulting from keyword assists)
Vendor Level Attribution
Looking beyond channel level, we use the same approach to assess the performance of each media buy. Shown below is a sample report showing the cost per action for each media vendor, both last-click and attributable. As shown, some media buys can appear to be very poor performers on a last-click basis, but are in fact very effective for creating demand that is subsequently satisfied through other channels.
Short-tail keywords (category terms, product terms, etc.) often play “assist” roles in the customer engagement process. Just as it’s important to know which display ads precede visits and conversions, assist keywords should also be identified. In many cases, assist keywords may perform poorly on a last-click basis, but perform very well in an attribution report.
The Business Case for Attribution
Attribution is more than just a buzzword – it is an essential part of campaign measurement and a requirement for optimizing media spend. As illustrated below, moving “loser” budgets to the “winning” vendors can produce a dramatic improvement in revenue and return on spend.
Beyond the improvement in media efficiency and ROS, the economic benefits also accrue to:
As an industry, we have to do better. We can’t use yesterday’s tools to measure tomorrow’s media. Attribution should no longer be an aspirational goal, but rather a key part of your 2011 digital marketing strategy. The economic returns are compelling and there are numerous vendors (including us!) who would be happy to assist you in taking a more holistic approach to digital media measurement and optimization.
As always, comments are encouraged. And please feel free to share!