ROI Solutions | Maximizing Marketing ROI

By Jeff Gentry and Logan Flatt, CFA

As a follow-up to ROI Solutions’ article, “Flaws and All: Viral Marketing and the Challenge of Measuring Viral Consumer Behavior,” we introduce here our Best Practices Checklist for Viral Marketing Using Email and Mobile Messaging. We created our Best Practices Checklist to ensure that our clients follow practices that encourage and facilitate “viral” consumer activity but also help mitigate omnipresent economic and legal risks inherent to viral marketing when designing interactive marketing campaigns that feature viral marketing as a tactic. We hope you find this tool helpful as you consider using viral marketing as a component of an interactive marketing campaign you may be designing or contemplating.

It must be noted that many of the best practices identified in our Checklist were originally identified in the 2006 ROI Solutions white paper, “Top Ten Legal Pitfalls of Viral Email Marketing,” by Jeff Gentry, Senior Marketing Strategist at ROI Solutions. At the time, the white paper was made available to our clients as well as internal staff at Tribal DDB Worldwide.

© 2008 Tribal DDB Worldwide. All rights reserved.

Bookmark to:

By Logan Flatt, CFA

Recently, one of ROI Solutions’ clients approached us excitedly about the potential of “viral marketing” and asked us to consider it during our development of the client’s 2008 interactive marketing plan. We, of course, happily agreed to do so.

At the outset of the plan development process, ROI Solutions and the client’s planning team jointly defined two key terms:

Viral, adj.: exhibiting large-scale, word-of-mouth communications among consumers.

Viral Marketing, n.: the facilitation of traditional word-of-mouth marketing among consumers by means of highly-scalable interactive communication tools, such as email, mobile messaging, and social networks.

By the end of the plan development process, our client arrived at three important conclusions about viral marketing. The sections that follow identify and describe some of the short-comings of viral marketing that our client considered to arrive at these conclusions as well as to make strategic choices about the role of viral marketing in the client’s 2008 interactive marketing plan.

Lack of Omniscience: The Inherent Flaw in Measuring Viral Consumer Behavior

During the planning process, ROI Solutions noted to our client that viral consumer behavior can occur almost anywhere that consumers use interactive technologies. However, we emphasized to our client that viral consumer behavior can only be measured and recorded when a predefined measurement solution is in place before any intended viral consumer behavior starts. Such a measurement solution typically takes the form of a database-driven application that must be designed, developed, tested, and implemented to record viral consumer behavior in the database before the start of an interactive marketing campaign. An example application for viral consumer behavior is a “Forward-to-a-Friend” email application that our client might make available on a branded micro site to enable consumers to promote a sweepstakes among them.

With the implementation of a predefined measurement solution required to record viral consumer behavior, it follows then that any and all viral consumer behavior occurring outside of the predefined measurement solution, or before its implementation, will go unmeasured and unreported. Continuing the example from above, if a consumer informs one or more consumers about our client’s sweepstakes via email but does so using Outlook, Thunderbird, Entourage, or some other desktop email program instead of our client’s “Forward-to-a-Friend” email application, that consumer’s viral behavior will go unrecorded and unreported by our client’s “Forward-to-a-Friend” email application. In this manner, the actual viral consumer behavior that occurs in total during our client’s campaign is not fully reflected in the measured and reported results that the client receives and considers for performance evaluation.

When it is known or believed that any viral consumer behavior occurred outside of a predefined measurement solution, the viral consumer behavior measured and reported by the predefined measurement solution is understated. In other words, more viral consumer behavior likely occurred but a marketer could not measure it or report it (perhaps only an omniscient being can measure and report all viral consumer behavior that occurred). Any management decisions to be made based on viral metrics must take into account that such metrics understate the total viral consumer behavior that occurred. In this manner, any reported viral metric serves as a lower bound on actual viral consumer behavior – e.g., “Consumers sent at least 1,329,538 viral email messages among them over the course of the six-week campaign.” ROI Solutions believes it is important that our clients appreciate this nuance about reported viral marketing results before they make substantive decisions regarding a marketing campaign or program featuring viral marketing as a tactic.

Two Lower Bound Measures of Viral Consumer Behavior

Most viral consumer behavior occurs through email or mobile messages. In such cases, viral consumer behavior features a “sender” who “distributes” a message to at least one “recipient” of that message.

Below, ROI Solutions defines two common performance indicators for viral consumer behavior facilitated through email messaging:

• Viral Email Distribution Counts
  The number of emails sent via a Forward-To-A-Friend application during a specific time period

• Viral Email Sender Counts
  The number of consumers who used a Forward-To-A-Friend application to send emails to other consumers during a specific time period.

Below, ROI Solutions defines two common performance indicators for viral consumer behavior facilitated through mobile messaging:

• Viral Mobile Distribution Counts
  The number of mobile messages sent via a Forward-To-A-Friend application during a specific time period.

• Viral Mobile Sender Counts
  The number of consumers who used a Forward-To-A-Friend application to send mobile messages to other consumers during a specific time period.

In the case of social networking (insofar as it stands today), measurable viral consumer behavior takes place largely through “widgets”:

Widget, n.: A snippet of code that, once installed in a consumer’s profile page on a social network, transforms into a value-added application that provides benefits to the consumer and/or to visitors to the consumer’s profile page.

Widgets become viral when a consumer in a social network decides to become a “sender” who “distributes” a widget’s “invite” to at least one other consumer, or “recipient,” in his or her social network. The invite gives the recipient the opportunity to install the same widget in his or her own profile page. Of course, the recipient may choose to reject the invitation to install the widget; nonetheless, viral consumer behavior has occurred – the sender distributed the invite to the recipient.

Below, ROI Solutions defines two common performance indicators for viral consumer behavior facilitated through social networking:

• Widget Invite Distribution Counts
  The number of invites sent via a widget’s Invite-A-Friend feature during a specific time period.

• Widget Invite Sender Counts
  The number of consumers who used a widget’s Invite-A-Friend feature during a specific time period.

For our client’s 2008 plan, ROI Solutions recommended that our client use two consolidated measures of viral consumer behavior, Viral Impression Counts and Viral Sender Counts. The Viral Impression Counts metric simply reflects the sum of all distribution counts that can be recorded and reported by our client’s predefined measurement solutions during a given period:

Viral Impression Counts
= Viral Email Distribution Counts + Viral Mobile Distribution Counts + Widget Invite Distribution Counts

The Viral Impression Counts metric serves as the measurable lower bound for the number of word-of-mouth communications sent by consumers to other consumers as part of our client’s interactive marketing campaigns. Clearly, other types of word-of-mouth distribution counts could be added to this sum, as warranted.

Similarly, the Viral Sender Counts metric simply reflects the sum of all sender counts that can be recorded and reported by our client’s predefined measurement solutions during a given period:

Viral Sender Counts
= Viral Email Sender Counts + Viral Mobile Sender Counts + Widget Invite Sender Counts

The Viral Sender Counts metric serves as the measurable lower bound for the number of senders of word-of-mouth communications as part of our client’s interactive marketing campaigns. Clearly, other types of word-of-mouth sender counts could be added to this sum, as warranted.

One Client’s Conclusions and Choices Regarding Viral Marketing

By the end of the plan development process, ROI Solutions’ client arrived at three important conclusions about viral marketing:

1. Viral marketing is not a marketing strategy in and of itself; rather, viral marketing is simply a tactic: it is one of several possible means to achieve the desired end (e.g., sales, leads, lift in brand awareness, opt-ins, registrations, etc.) of any campaign type (e.g., Direct Response, Brand Awareness, Brand Response, etc.)
2. Viral marketing can be engineered into the design of most interactive marketing campaigns regardless of campaign type
3. Viral consumer behavior can only be measured and recorded for performance evaluation purposes if a predefined measurement solution has been engineered into the design of the interactive marketing campaign, and even then, not all viral consumer behavior will be measured and recorded

Based on these conclusions, our client made two important choices about how to incorporate viral marketing into its 2008 interactive marketing plan. First, the client chose to use viral marketing as a tactic only – interactive marketing’s ability to facilitate viral communications among consumers is powerful, but its power is only useful if it helps generate more important marketing outcomes, such as sales transactions, qualified leads, opt-ins, registrations, and/or a lift in brand awareness. Second, our client chose to not elevate any measures of viral consumer behavior, such as Viral Impression Counts and Viral Sender Counts, to the level of Key Performance Indicator (KPI) for any of its interactive marketing campaigns in 2008; the client understood that marketing outcomes of a higher order than viral metrics make more appropriate KPIs for interactive marketing campaigns.

ROI Solutions seconds our client’s conclusions and choices regarding viral marketing. Yes, much viral consumer behavior goes unmeasured and unreported, leaving the interactive marketer with only a measured lower bound to a complete viral marketing story. Still, this inherent flaw in the measurement and reporting of viral marketing results is acceptable: while it remains an important tactic in a marketer’s bag of interactive marketing tricks, viral marketing is simply a means to an end, not the end itself.

© 2008 Tribal DDB Worldwide. All rights reserved.

Bookmark to:

^{Copyright 2007 Eric Bergman}

Interview of Logan Flatt, CFA, President of ROI Solutions, for the podcast, “Media Relations Matters” by Eric Bergman, principal of Bergman & Associates in Toronto, Canada. In this podcast, as a “measurer of marketing,” Logan Flatt provides his insights into the communication measurement process and his view that the ultimate goal is “to measure changes in behavior.”

Bookmark to:

By Logan Flatt and Cindy Wells

Recently, a client asked us, “How do you define effective reach?”

Good question! Having studied advertising theory at university, we generally know of effective reach as an academic term with greater application to offline media and, perhaps, greater use by offline media agencies. However, we are practitioners working day-to-day alongside corporate clients facing online media challenges. Over time, we’ve come to learn that, out of necessity, what makes sense in practice often trumps what makes sense in theory, particularly when trying to apply offline theories to online practicalities.

It should come as no surprise then that we provided our client with a rather non-academic response to her question: we define effective reach fully within the context of a client’s stated objective for the reach of a particular media campaign. If a campaign’s stated reach objective was achieved, the campaign’s reach was effective. In contrast, if a campaign’s stated reach objective was not achieved, the campaign’s reach was ineffective. It’s hard to get any more practical than that.

In practice, during the planning phase of a campaign, one could establish a target reach in the stated reach objective that predetermines the inflexion point at which the actual reach achieved by the campaign can be deemed effective or ineffective. At the end of the campaign (or, perhaps even during the campaign’s execution) one could critically evaluate the campaign by comparing the reach it achieved against the target reach built into the campaign’s stated reach objective.

For example, we could establish a target reach of 53 million U.S. individuals for a particular campaign’s reach objective, subject to budget constraints. If after the campaign ends we find that the campaign achieved a reach equal to or greater than 53 million U.S. individuals, then it could be said that the campaign’s reach was effective, or that the campaign had an effective reach. However, if the campaign achieved a reach less than 53 million U.S. individuals, then it could be said that the campaign’s reach was ineffective, or that the campaign had an ineffective reach.

Let’s take a look at another example. A client’s stated objective for a media campaign might be to maximize reach, subject to budget constraints. In that context, the reach achieved by the campaign would be deemed effective only when said reach was maximized given the budget employed. If the same budget could have achieved greater reach (i.e., through better strategy, better tactics, better negotiations, more reach-productive channels, etc.), then the reach achieved by the campaign could be deemed ineffective. Similarly, if a smaller budget could have achieved the same reach (i.e., through better strategy, better tactics, better negotiations, more cost-efficient channels, etc.), then the reach achieved by the campaign could be deemed ineffective. In both cases, the reach achieved was suboptimal to the client’s maximization objective and therefore would be difficult to deem effective in that context.

Clearly, our definition of effective reach — and, for contrast, ineffective reach — favors practical application over academic theory. Alas, such are the required tools of our trade. We note that, by our definitions, the difference between a campaign having an effective reach or ineffective reach depends heavily on the quality of the media strategists allocating a client’s scarce media dollars across competing media alternatives in pursuit of the client’s stated reach objective. At ROI Solutions, we prefer to develop and use definitions, tools, and techniques that best fit our clients’ immediate, “real world” conditions. We believe our practical approach, as exhibited in our definition of effective reach, makes our media strategists highly effective allocators of our client’s media dollars.

Bookmark to:

By Cindy Wells

Recently, one of our clients asked ROI Solutions to describe the factors it uses to evaluate new, alternative media opportunities as they arise. Our response was simple: we identify, evaluate, and assess the degree to which each new opportunity represents a sufficiently compelling solution to each client’s unique marketing needs.

In general, evaluating and assessing such a broad range of new opportunities requires a laser-like focus on four evaluation factors and, in our opinion, their relative importance weightings:

Clearly, the new, alternative media vehicle’s alignment with client marketing objectives, degree of target audience impact, and cost efficiency are, in total, critical evaluation parameters.

Important to note, however, is the risk of underperformance — the risk that a client’s scarce marketing capital goes wasted on a new, alternative media vehicle that does not generate results as compelling as a well-established media vehicle with a proven track record would have. After all, the opportunity cost of poor performance is real. We work with our clients to mitigate that risk, particularly in the face of intense pressure to allocate marketing capital to new media vehicles that are fashionable and trendy at the time. Limited testing of a new, alternative media vehicle with only a small amount (i.e., 1% to 10%) of a client’s media budget is a prudent way to mitigate the risk of underperformance.

Even if a new opportunity initially does not score well along one or more evaluation factors, ROI Solutions does not ignore the opportunity outright. We constantly monitor new media opportunities as they evolve over time to determine if they warrant re-evaluation and, possibly, further investigation through additional testing.

© 2007 Tribal DDB Worldwide. All rights reserved.

Bookmark to:

By Steve Fedorko, Ph.D.

Everything that can be counted does not necessarily count; everything that counts cannot necessarily be counted. – Albert Einstein

Unlike traditional one-way broadcasts, interactive channels, such as web, email, and wireless, provide better opportunities to directly measure consumer behavior and the myriad effects of marketing campaigns. In fact, there are so many things that can be measured in the interactive space that it becomes necessary to separate the wheat from the chaff. Those who seek insight and knowledge from interactive metrics would be well served to remember that just being able to count something doesn’t mean it is valid, valuable, or indicative of some other relevant measure, like customer loyalty, likeliness to purchase, or product sales. Let’s look at a few current exemplars from across the board.

The Good

Visits – A visit is basically a browser user session. A site is reached, that site is left, and that’s one visit. A count of visits would be the number of visits all consumers made to a site during a specific time period. Visits are a meaningful interactive metric because they are a count of users actually initiating the serving of specific domain pages to their browsers. It might be ten visitors making a single trip to a site or one visitor making ten trips, but they were voluntary, user-initiated trips, which intimate a basic interest in the site content or brand. As well, this metric scales well and is comparable across sites or campaigns, much more so than, say, counts of wallpaper downloads or sweepstakes entries.

Time on site measures – The two predominant measures of time on site include average visit duration and cumulative consumer time. Average visit duration simply reflects the time from the initiation of a visit until the visit ends, while cumulative consumer time is the aggregate of visit durations across all visits within a specified time frame. Like visits, we can generally assume that more time means more interest. (One could argue that something like poor usability could lead to longer visits, but users probably tend to leave sites with poor usability sooner so duration is shorter anyway.) Although not an exact match, cumulative consumer time can be compared to cumulative time consumers spend with other channels, such as television, and costs per hour and effects can be evaluated in a preliminary manner.

Consumer relationship measures – Metrics like consumer registrations, opt-ins, and RSS subscriptions demonstrate a level of interest beyond simply visiting a site or spending time there. Consumers provide information about themselves and in many cases also give permission to be contacted or have content “pushed” to them. Any of these measures signal the first and necessary step in the transition from site visitor to identified consumer. Interactive’s ability to create, maintain, and grow customer relationships is a positive differentiator over television, radio, print, and outdoor.

Cross-channel measures – Interactive metrics which extend across multiple channels bring particular benefit as they adopt the proved value of the other measures and also provide direct opportunity for comparison and evaluation. Scheduling vehicle test drives and inquiry-to-lead ratios are just two examples of effective cross-channel measures. Automobile dealerships know that they will generally make a sale for every certain number of test drives. Consumers scheduling test drives interactive follow a similar predictive model so these numbers can be used to evaluate the effectiveness of respective site designs as well as provide the data necessary to compute ROI and compare different investments in other test-drive producing channels. Inquiry-to-lead ratios (inquiries by interactive visitors who become qualified customer prospects) are established metrics which often translate interactive and provide comparable benefit in areas like sales and recruitment. Cross channel measures are perhaps the best metrics for interactive in that they are directly comparable to traditional channels and leave little doubt as to channel effectiveness.

The Bad

Hits – A “hit” is a count of any web file served to a browser, including an HTML page, an image, or a sound. In the early days of the World Wide Web, with simple pages and slow connection speeds, hits were at least some measure of traffic to a site, but everyone quickly saw that a visit to a site could mean 10 hits or 100 simply depending upon how many image or sound files were used to create each web page. Although hits can be measured quite accurately, they never were a good measure and their irrelevance has not changed. If someone talks to you about hits today, that’s a red flag.

Page views – A page view is generally a browser’s request to serve another “page” of a website. Another “page” could be branching to a fully different web page, but it could also be playing a sound effect, launching a small pop-up, or even just changing some text or graphic of the current page. Since a “view” depends on how the site was constructed, page view counts turn out to be more a measure of site architecture than consumer interest in a brand or likeliness to purchase. Page views might be occasionally interesting but only in limited circumstances; for the most part they have little relevance in judging consumer behavior.

IP-based geography – If you could tell the geographic location of the visitors to your website, you could draw conclusions about what kind of audiences you were attracting and where your site was popular. But tracking the geographic location of visitors is more likely a measure of the ISP locations providing connectivity to these visitors than it is an accurate indication of where the visitors actually are when contacting your site. Location measures of visitors are derived from visitors’ IP addresses, and these are actually reflective of the locations of the internet service providers (ISPs). ISPs generally serve their surrounding areas, but these areas can get very large even covering many states (e.g., AOL servers in Chantilly, VA) so very specific location information about visitors is not readily available. At a global level, differentiation between countries is probably reasonable, but accuracy at the state or city level is just not reliable. How else would the top ten city lists of several national high-traffic sites consistently include Chantilly, VA, Marina Del Rey, CA, and Plano, TX?

The Ugly

Visitors – How could a count of visitors be an “ugly” interactive metric? Although everyone would want to know how many actual people visit their site, a count of unique visitors is really a count of saved cookies on individual computers. Some computers have several users (mom, dad, daughter, son), and some people use more than one computer (home, work, school, folk’s house, friend’s houses). More and more people now delete or refuse cookies. So, estimating a count of actual people from a count of “unique visitors” (cookies on a computer) could be off by 50% or more, high or low. Visitor counts are additionally confounded depending upon the selected time periods of measurement. Measuring weekly, a unique visitor (i.e., computer cookie) who visits a certain site once every week would count as 52 “visitors” in year; however, the exact same data measured monthly would count as 12 “visitors” instead. Measure once a year and you’d have one unique visitor. With these potential variance swings, better to count visits than visitors.

Average visits per unique visitor – Knowing how many visits unique visitors made to a site sounds good, but combining a good metric (visits) and an ugly metric (unique visitors) doesn’t beautify the ugly; it just compromises the good.

Sweepstakes entries – Sweepstakes are as popular online as they are in the offline world, and probably just as ineffective. Counting sweepstakes entries may be indicative of interest in the prizes, but maybe not so much brand awareness, likelihood to purchase, or even reach to the target audience. Some contend that if it produces visits and entries, then it raises brand awareness. This might be occasionally true for certain contests, but given the high costs of sweepstakes (site production, prizes, legal, fulfillment, etc.) and the plethora of untargeted “sweeps-junkies” on the web – and the sites which support them – there are more cost-effective ways to increase brand awareness online.

Cross-campaign comparisons – It’s a natural and tempting proposition to compare results across two or more sites. But for this type of benchmarking to be valid it is important to carefully analyze and match the multiple salient variables which would make the sites similar in the ways you would like to draw conclusions about their respective performances. In order to assess the business impact of two or more sites, the specific features of the site need to be similar enough to isolate one or two key variables which could account for any difference that might be observed in the business outcomes. For example, it would be difficult to accurately explain the causes of the observed business results (e.g., online visitor-scheduled test drives) when comparing two very different sites – a large content-rich site with sweepstakes and TV media launched in the summer compared with a special promotion site for an end-of-year clearance with celebrity tie-in and rebate coupons for loyalty program customers. In science-speak, there are too many different independent variables operating to permit conclusive, or even meaningful, insights about what caused any observed differences in the dependent variable (i.e., online visitor-scheduled test drives). It’s the classic apples-to-apples argument and it’s as important here as it is anywhere else.

Every interactive case brings its own special characteristics and objectives, and business measures should be matched to those needs. Interactive measurement will continue to evolve just as radically as interactive channels themselves, but some basic guidelines for selecting and analyzing the right business metrics are emerging. A few current exemplars were discussed above – some better than others, some better left behind.

© 2007 Tribal DDB Worldwide. All rights reserved.

Bookmark to:

By Logan Flatt, CFA

Recently, Advertising Age announced¹ the publication of two research studies centered on measuring and evaluating the performance of Chief Marketing Officers and other marketing executives (CMOs) at some of the world’s largest for-profit corporations. Based on the study authors’ short comments in the announcement, I question the authors’ conclusions to the studies and offer an alternate perspective on those conclusions.

A Tale of Two Studies

The first study², an academic inquiry into the CMO’s effect on a company’s financial performance, covered 167 for-profit companies over a five-year period. The study concludes that “CMOs on top management teams don’t have any effect on a company’s financial performance.” However, the authors quickly caveat the conclusion by adding, “[our] study is limited because it focuses on financial-performance metrics, such as sales growth and profitability, and not brand equity.”

The second study³, the collaborative effort of a management consulting firm and an advertising industry association, relied on insights gained from “15 incisive and revealing interviews with former and current marketing leaders at household-name companies” to conclude that “measuring CMO performance based on financial performance alone is a mistake.” One co-author of the study declares, “Financial metrics alone do not define CMO performance.” Continuing, he proclaims, “A universal panacea to making marketing accountable doesn’t exist.”

CMO Performance Metrics Must Reflect the Capitalist Imperative

As a plain-speaking Texan, I respond to the authors’ conclusions and proclamations with a drawled “Hogwash!” Why do I feel they are wrong? The authors’ conclusions do not reflect the reality that a for-profit CMO must adhere at all times to the Capitalist Imperative – the prime directive from shareholders to “use our money to make more money.” Under the Capitalist Imperative, financial metrics such as sales, costs, profits, cash flows, and ROI are the only relevant metrics to use when measuring and evaluating CMO performance.

In contrast, traditional marketing metrics such as “lift in brand awareness,” “intent to purchase,” and “brand equity” collapse under the weight of the Capitalist Imperative because they do not measure a CMO’s direct effect on a company’s ongoing financial performance. At best, traditional marketing metrics merely suggest – in vague terms – a CMO’s possible effect on company financial performance. At worst, traditional marketing metrics distract from – or can be used to mask – a CMO’s inability to use shareholders’ money to make more money. Not surprisingly, traditional marketing metrics are of little interest or use to most decision makers outside of the Marketing department. It is no wonder then why CEOs and CFOs do not report traditional marketing metrics in their quarterly and annual reports to shareholders.

Let’s Give Shareholders CMO Performance Metrics They Deserve

Shareholders need to understand and appreciate a CMO’s contribution to company financial performance. A CMO often has discretionary spending authority over many millions of shareholder dollars for company marketing initiatives each year. Therefore, any evaluation of a CMO’s performance must be linked inextricably to financial metrics that reflect whether or not he or she spends those dollars in ways that create value for shareholders.

Yes, measuring and evaluating CMO performance using only financial metrics is a difficult task. It requires a good number of assumptions, estimates, and fancy mathematics to get the job done. Still, hard work and imperfection are not good excuses for simply throwing up one’s hands and declaring that the job cannot or should not be done. Shareholders deserve better than that. They deserve to know a CMO’s contribution to company financial performance in the terms most relevant to shareholders – through financial metrics alone.

************

References

¹“CMOs Rapped for Having Zero Impact on Sales: Study Shows Difficulty in Measuring Short-Term Value of C-Suite Position,” Mya Frazier, Advertising Age, July 9, 2007.

²“Chief Marketing Officers: A Study of their Presence in Firms’ Top Management Teams,” Pravin Nath and Vijay Mahajan, Journal of Marketing, forthcoming in January 2008.

³“CMO Thought Leaders: The Rise of the Strategic Marketer,” Gregor Harter, Edward Landry, and Andrew Tipping, edited by Geoffrey Precourt, strategy+business, published 2007.

© 2007 Tribal DDB Worldwide. All rights reserved.

Bookmark to: