Historically, account managers contact customers to confirm the renewal terms and timing.  The practice stems from the on-premise product world where customers could continue to receive value whether their maintenance contracts had been paid.  As you can imagine, the practice has continued into the subscription economy, but it is not necessary.  In the subscription economy, a customer loses access if their renewal is not processed.  If a customer values the product or service, they will be sure to renew before losing access.  So the responsibility and effort belongs to the customer in these cases.  Using automatic renewals eliminates this cost inefficiency.

In the subscription economy, the old method of renewal also introduces a drag on revenue.  When an account manager contacts the customer, the subsequent interaction often creates negotiations.  Customers use these instances to negotiate away standardized price increases or to achieve deeper discounts.  Using automatic renewals eliminates the majority of these negotiations and consequently transforms the revenue drag into revenue growth.

However, automatic renewals can hurt revenues if not used appropriately.  For example, automatic renewals can erode revenue if the customer is not getting a good return on investment.  The customer may be looking at the renewal as a chance to terminate use of the product or service.  An automatic renewal in this case may force the cancellation. For customers getting above average return on investment, automatic can stunt up-sell opportunities.  The renewal represents an opportunity to increase prices to match customer value.  So targeting automatic renewals to the right customers is critical.

When automatic renewals are targeted effectively to the right customers, a company can lower costs and increase revenues without risking cancellations or stunting up-sells.  The following example illustrates the impact.  Our case example is a human resources management solution which had one million dollars of subscriptions up for renewal.  In the company’s customer base, approximately sixty percent of the renewals were predicted to be candidates for automatic renewal.  By placing those accounts into an automatic renewal process that consisted of issuing an renewal agreement sixty days prior to renewal and a follow-on reminder thirty days prior, they were able to increase the renewal yield from an eighty-six percent to ninety-four percent.  The eight percent increase in automatic renewals had an overall revenue yield increase of four point eight percent which is broken out in the infographic.

The benefit of increased revenue yield is compounded by the fact that account managers eliminated sixty percent of the contacts and preparation time.  Account managers let normal customer success management and service delivery to do their jobs in creating satisfied customers.  Instead, they invested their time where there is more revenue impact – up-selling and supporting retention efforts.  The results are that the other forty percent of customer relationships grow more quickly; and the overall renewal yields easily increase beyond the four point eight percent.

While the example provided is a B2B example, the same logic can be applied to consumer subscriptions as well.  Targeting price increases and rate plan changes at scale requires the same kind of predictive analytics to know who should get what pricing or needs what intervention.

The take away is that targeting automatic renewals to the right customers can be a revenue accelerator and create additionally account management productivity.  Without predictive analytics, automatic renewals stunt up-sells and increase churn which is why so many organizations still rely on manual procedures which hampers growth.

While these scores provide insights on customer satisfaction, they are limited in their ability to prevent churn. A real example provides a case in point, but first a little more information about the NPS.

The NPS is a customer loyalty metric developed by Fred Reichheld, Bain & Company, and Satmetrix. The 2003 Harvard Business Review article, “One Number You Need to Grow,” introduced the concept. NPS measures the relative amount of customers who are promoters of a particular brand or company (i.e., fans) versus those who are detractors by measuring the difference between the two. NPS can be as low as −100 percent (everybody is a detractor) or as high as +100 percent (everybody is a promoter). By knowing this score and tracking changes, each department from marketing to products to sales to support can understand how well their products and services create customer satisfaction.

The Good: The NPS is a valuable metric for gauging sentiment of the customer base — we use it at Scout Analytics. Because the score is derived from a survey, NPS provides sentiment via a sampling of an organization’s customer base, typically with a 10 percent to 15 percent response rate.

The Bad: While sampling provides a good representation of the whole base, it does not provide specific insight into every individual customer. In fact, 85 percent or more customers are not represented by NPS.

The Ugly: For customer success organizations wanting to leverage NPS to prevent churn, the lack of coverage (i.e., 85 percent of customers without an NPS) is a real issue. Investments in time, money, and other resources to implement an NPS will quickly bump into the laws of diminishing returns in terms of predicting and preventing churn.

An Actual Case in Point
This case illustrates not only the coverage issue, but also a second concern regarding NPS accuracy in predicting churn. A midsize SaaS firm in human resources management was using the NPS as a predictive factor to determine at-risk customers. When an audit was completed to determine predictive accuracy, the company was surprised to find that more than one-third of its so-called “promoters” churned. Likewise, only 25 percent of their detractors churned, including those that received no intervention.

Peeling back the details on its NPS revealed the mismatch between NPS and renewals. The NPS survey had been sent to all customer contacts (i.e., users and buyers of the SaaS service). While the users generally had more promoters, the buyers had more detractors. In other words, the users liked using the service while the buyers were not happy with the return on investment. In fact, the composite NPS (i.e., NPS from all users and buyers) was +9.4 percent, but among buyers only it was -20 percent — a low score. And because the buyers, not the users, are responsible for the renewal decision, the use of composite NPS provided inaccurate predictions. The composite NPS predicted churn correctly only 23 percent of the time. Given its response rate was 12.3 percent, its customer success teams had low customer coverage and predictive accuracy.

How Do You Avoid the Pitfall?
Understanding customer coverage and predictive accuracy before using a metric for churn prediction is important. If the metric used has low customer coverage and predictive accuracy, then customer success teams will constantly be in fire drills. If the metric has high coverage and low accuracy, then the customer success team will have many false alarms. Likewise, high accuracy and low coverage will be surprised by unanticipated churn. Only with high accuracy and high coverage can you get to efficient and effective churn prevention.

The following infographic summarizes the challenges of NPS for predicting and preventing churn for this specific SaaS company.

NPS remains a valuable and useful metric for companies to know and grow; however, it rarely provides sufficient actionable data for customer success teams either because of coverage or accuracy.  Rather, customer success teams should view NPS as a supporting metric.  Our company uses NPS as a critical component of understanding how we can improve. That said, we focus on usage and ROI metrics for managing customer success.

Usage data is the new fulcrum for managing profitability of customer relationships. It might seem like an obvious statement, but the only way to gauge customer success is based on whether customers actually receive value from use of your product. And without visibility into critical milestones of the customer lifecycle, such as on-boarding and adoption, unanticipated churn and customer retention fire drills become problems too great to ignore.

What Is Usage Data?
Usage data can be defined simply as “data about usage.” It is the record of customer activity occurring at any given time. For example, usage data can include a page view, a file download, a transaction performed, or an article read. Before so many products and services turned to digital, usage data was not a factor in determining how much value a customer was getting out of a subscription-based service. But with the shift to digital, massive volumes of usage data are now being generated daily, and service providers now have the opportunity to measure and monitor content consumption,use of features, and frequency of use. The companies that connect usage data to customer success management, product management, marketing, and sales are the ones achieving profitability.

Usage data can answer such questions as: Have active users gone dormant? Is a customer’s usage above or below the average? Is a customer using all their subscriptions? Has the customer usage levels dropped off?

Usage data leads to metrics with high coverage and high accuracy for predicting and preventing churn, and is the key to having an efficient and effective customer success team.

What are these facets and how do they affect your organization?

Customer coverage is the percentage measurement of the customer base that the scoring rule provides insight.  For example if a rule is implemented based on survey response there will be low coverage. Typically, 20 percent is considered an awesome response rate for a survey.  Most survey response rates are in the 10-15 percent ranged. The implication is that more than 80 percent of you customers will not respond to the survey which means that scoring rule based on survey responses has low customer coverage.   Creating rules around other metrics such as number of open support cases or attendance at customer events can fall into the low coverage category.  Rules with high customer coverage include demographic, firmographic, purchase history, and usage metrics.

Predictive accuracy is the percentage measurement of a rule’s true versus false predictions.  In the case of churn, if a scoring rule predicts someone to be a non-churner and they do in fact churn, that is a false prediction.  Likewise another false prediction would be if the rule predicts a churner and without any intervention from customer success, the churner renews.  True predictions are of course the opposite of these.  The percentage of true predictions versus the total provides predictive accuracy.

So how do these metrics provide help discover ROI for a customer success team?

Customer success teams are most efficient and effective if they intervene only when needed – i.e., if a customer is at risk.  Interventions with customers not at risk are not only an unwarranted expense but also represent opportunity costs of assisting the actual at-risk customers.  Therefore, low predictive accuracy actually lowers customer success organization’s efficiency and effectiveness.  And low customer coverage, means missed opportunity.

The dynamics of customer coverage and predictive accuracy are shown in the following graphic.  If the metric used has low customer coverage and predictive accuracy, customer success teams will be constantly reacting to fire drills in order to save customers.  If the metric has high coverage and low accuracy, the customer success team will have lots of false alarms resulting in wasted effort and opportunity costs.  Likewise high accuracy and low coverage will result in unanticipated churn.  Only with high accuracy and high coverage can you get to efficient and effective churn prevention.

The take away is that predictive analytics is a powerful tool for creating an efficient and effective customer success organization.  With that said, predictive analytics developed from experience and intuition without validation can actually have the opposite effect.

Step 1: Measure the Gap
To evaluate if you have a consumption gap, you need to measure and quantify it. That means you need to take inventory on your product’s capabilities and how much a particular customer uses them.  For offering, the most straightforward measurement is price. The more features a product provides, the more a business can and will charge. Customer usage can be directly measured using a variety of metrics associated with consumption, such as number of transactions, number of data look-ups, number of active users, etc.  The gap is therefore measured as the difference between usage (i.e., value received) and subscription price.

But since usage is not a monetary metric like price, how do you calculate the difference? The answer is yield. Yield is a calculation that combines price and usage into a ratio to provide revenue per unit of customer fulfillment (e.g., revenue per seat mile in the airline industry).  Yield then provides interesting insights. Figure 1 shows a yield chart where usage is measured on the horizontal axis in units of usage and price is measured on the vertical axis in dollars.  Any customer’s subscription can be mapped into this chart.  As shown by Subscription 1 when yield is too high (i.e., low usage compared to high price), the subscription is overpriced compared to customer usage meaning there is a consumption gap and renewal risk.  As shown by Subscription 2 when yield is too low (i.e., high usage compared to low price), the subscription is underpriced compared to customer usage meaning there is an excess rather than a gap in value and a revenue opportunity is available.

The following is a real example from a SaaS company that is a customer. The data is blinded but shows how the consumption gap impacts revenue and how yield is a good predictor.  In this SaaS example, the most important usage metric is active users.  Other SaaS companies have used ‘number of transactions’ or ‘number of artifacts,’ but in this case the right metric was ‘number of active users.’  Figure 2 illustrates the correlation of customer renewal behavior to yield.  Revenue growth came when yield per active user was about $1,500 (i.e., customer is getting good value and wants more), and revenue churn occurred when the yield went toward $3,000 (i.e., customer was not getting good value for the price paid).

Step 2: Mind the Gap
Making sure the yield from each subscription is “just right” is the basis of recurring revenue management, and is essential to ensuring success in the Subscription Economy.  Recognizing which customer needs help increasing usage can curtail revenue churn.  Knowing which customer is an up-sell candidate creates revenue growth.  Overall, understanding yield allows a subscription-based business to offer the right subscription, to the right customer, at the right price, at the right time.  Realizing yield management enables subscription based companies to avoid the consumption gap and maximize customer lifetime value.

What is revenue management?

Revenue management is common practice in the distribution-centric Transaction Economy.  The goal is to maximize revenue and profits by pricing products to match customer demand.  Revenue management is pervasive in such industries as airlines, hotel rooms, surgery, advertising, retail, media and rental cars.  For example, airlines offer a passenger a seat between two cities defined by departure time, legroom, seat width, and associated service.  Because the product, in this case a seat, is both standardized in terms of customer fulfillment and limited in inventory, the airline can forecast demand at specific prices from different customer segments and manage seat availability to optimize revenue.  The airline can forecast demand for higher priced seats from business travelers that value last-minute bookings and sell the remaining inventory at a lower price to early purchasing leisure travelers who value cheap travel.  While the business traveler and the leisure traveler sitting next to each other expected the exact same product, each valued the trip differently and consequently paid a different price.  By selling the right standardized, inventory-constrained product to the right customer at the right price, the airline maximizes revenue and profit.

Why can’t traditional revenue management be used in the Subscription Economy?

Unfortunately, distribution-centric revenue management doesn’t work for the consumption-centric Subscription Economy.  For example, imagine if your cellular provider informed you that all the minutes of data transfer were sold out for the day and you could not buy anymore regardless of price?!  Or imagine if you wanted to sign up for a subscription and the provider said they were sold out?!  These are principles of the distribution-centric revenue management process.

The revenue management process is different in the Subscription Economy for two reasons as shown in the figure.  The first difference is that customer fulfillment is variable.  While each customer receives a standard subscription agreement, each of them will use the product differently.  Unlike the airline example where the airline defined customer fulfillment (i.e., a seat), in the Subscription Economy, customers define fulfillment based on their individual usage (e.g., amount of texting consumed in a cellular plan).  Customer demand can no longer be determined from purchase data alone.  Customer demand must be measured by usage data and purchase data together.

Second in the Subscription Economy, there is no limitation in inventory.  In other words, differentiated value between customer segments cannot be derived simply from product availability (i.e., inventory management).  Unlike airlines that create differentiated value and revenue based on managing inventory of a particular product package (i.e., a seat), in the Subscription Economy, differentiated value and revenue opportunities have to be created by providing differentiated product packaging (e.g., different combinations of minutes, text, and data in a cellular plan).  Rate plan management replaces inventory management for revenue optimization.

These differences highlight why revenue management in the Subscription Economy requires a new approach which the “use it or lose it” dynamic highlights the most. The “use it or lose it” dynamics states if the customer does not use your product at a level that matches the subscription agreement, the customer will cancel the subscription, and you’ll lose the revenue (A complete description of “use it or lose it” can be found here ).  So in the Subscription Economy, recurring revenue management (i.e., the ability to proactively manage the subscription revenue model) boils down to matching the right rate plan to the right customer usage at the right price.

Why is recurring revenue management required?

Just as revenue management was is a well-chronicled competitive advantage in Transaction Economy industries such as airlines, recurring revenue management will be a requirement for competitive advantage in the Subscription Economy.  Recurring revenue management is rapidly becoming a necessity rather than a nicety.  Companies that currently leverage recurring revenue management, are able to increase topline revenue 10-15% compared to their competition.  That statistic by itself makes recurring revenue management a required discipline for survival the Subscription Economy.

• Which audience members read the latest news on a particular topic?
• What news is the most popular for which audience members?
• Which audience members read a vendor guide or some other directory?
• Which audience members are researching information on a specific type of product?

Answering these questions allows media companies to behaviorally target editorial that drives audience engagement, to behaviorally target metering to increase registrations, and to behaviorally target offers to increase event attendance.  A behavioral database increases revenue – and that is why the result of the benchmark was so surprising.

A behavioral database tracks what each member of the audience does.  A behavioral database also allows media staff to search and answer questions such as those posed above.  How many articles did a particular audience member consume?  What type of articles? On what topic?  How frequently?  Is the audience member in market?

In the benchmark, my loose definition of “use” was that the media staff could search directly or through another staff member and the answer would be available within 24 hours.  The digital media brands in my benchmark specifically represented mainstream B2B publishers rather than the leading innovators that already have these capabilities.  Out of 20 digital media brands, none of them had a behavioral database.  It does not mean they didn’t have behavioral data, since almost everyone has weblogs.  It means they had no operational use of behavioral data for targeting within editorial, audience development, event marketing or advertising sales.

The odds of my benchmark reporting only anomalies are astronomically low – the same odds of flipping heads 20 times in a row.

If you are a digital media publisher, how would you answer that question?  Do you have a searchable, analyzable and reportable behavioral database?  Can you query a behavioral database on demand to find out which of your audience are “in-market” (e.g., reading product evaluation guides) for which products?

Managing the market to optimize revenue

Part of the reason for the lack of behavioral databases can be attributed to the lack of commercial offerings.  Other reasons come down to investment priorities in new channels such as mobile.  Whatever the reason, the lack of a behavioral database is one of digital media’s biggest missed opportunities.  Here’s why.

The primary role of digital media is as a marketplace to introduce buyers and sellers.  Of course there are other purposes, but as measured by the percentage of revenue from audience versus advertisers, the linkage of buyer to seller is the dominant role.  Consequently, a digital media brand has to be able to identify potential buyers for sellers.  Is a potential buyer defined by a title in a certain size company within a particular industry?  Maybe, but the truth is even if the advertiser’s target market was so tightly defined, only 10 percent of the titles are in market at any one time.  Therefore, managing who is in market and for what reason optimizes revenue.

Consequently, the better definition of a potential buyer is an audience member who exhibits buying intentions (i.e., behaviors).  With that definition, the only way to detect potential buyers is to make use of a behavioral database to analyze and understand behaviors within the buying funnel.

Now contrast the lack of behavioral databases in digital media with what advertisers are investing in:  marketing automation.  Marketers are constructing huge databases about their prospects and customers in order to assist in understanding buying behavior, architect customer acquisition funnels and optimize spending and revenues.

Marketers are becoming hugely data-driven, moving beyond simple cost-per-lead metrics into areas such as revenue attribution by funnel and marketing asset, conversion rates at each step in a funnel, and funnel performance over time.  And the more options marketers have to reach potential buyers (e.g., social, search, content marketing, press, events), the more behavioral data-driven they become.

To stay relevant in this world, digital media has to be more behavioral data-driven than the marketers themselves (i.e., their customers).  Digital media brands need to be able to quickly and easily identify funnels and buyer activity within their own media.  Only then can they segment their audience members to be in or out of market for different advertisers.  Only then can they help marketers construct high-performance funnels and earn high margins.

A digital media company should be able to answer the following questions for advertisers:

• What are the correlations between specific media topics and conversions?
• What types of content (e.g., white papers, evaluation guides, etc.) create more conversions than others?
• How many audience members are in market at any one time?

A digital media brand with a behaviorally data-driven approach creates a virtuous revenue cycle for itself.  The more a digital media brand knows about what its audience is in market for, the more the digital media brand knows how to service each audience member.  Guess what?  Audience engagement increases.  Audience size increases.  Digital media brand value grows with advertisers as conversion rates increase.  Up go the revenues.

Digital media companies definitely face huge challenges in developing profitable and sustainable revenue models, but the foundation to that model is a marketplace that benefits the buyer and seller.  Without understanding the correlation between media consumption behaviors and buying behaviors, digital media brands are missing the real opportunity.

On the other hand, if a customer’s usage drops below certain thresholds, the first thing that occurs is tier or seat churn, which reduces subscription revenue. Ultimately, continued drops in usage lead to product churn as the customer decides to use less of the service. Finally, customer churn occurs when usage falls so low that regardless of price the subscription is not justified.

So how much usage is enough to retain subscription revenue? To grow subscription revenue? To cross-sell other products? And at what point does usage decline create tier or seat churn? Product churn? Customer churn? And are the usage thresholds the same for every product and every rate plan? And finally, how do you measure usage: by active user, by artifacts created, by transactions, by report views? The only way to know the answers to these questions is to collect and then analyze usage data in the context of subscription data.

Usage data is simply the data recorded about usage. Its volume, variety, and velocity make it difficult to harness. The volume of usage data includes billions of user events to be analyzed for correlations and insights. The variety of usage data ranges from web events, mobile application events, call center activity, and other customer interaction points. And the velocity requires millions of new usage events being recorded and responded to every day. To complicate things more, usage data is generated and stored in systems that are separated from the subscription data, so attributing (i.e., linking) usage to the right rate plan, product, subscription, and customer can be tricky.

As part of Z-Finance Open, Scout Analytics automates that integration of usage and subscription data. Scout Link for Zuora links usage data managed by Scout Analytics with subscription data managed by Zuora. The result is predictive analytics utilized to drive revenue growth. The integration of the two sets of data allow correlations between usage and subscription retention, growth, and churn to be identified and quantified. Consequently, as usage data accumulates:

• Trial prospects can be segmented by those most likely to close
• At-risk customers can be identified early to prevent churn and retain revenue
• Renewals can be classified as cross-sells and up-sells to grow revenue
• Rate plan prices and packaging can be modified to optimize revenue

In other words, predictive analytics made possible by Z-Finance Open and Scout Analytics can be used to optimize subscription lifecycles and customer lifetime value. By linking usage data to subscription data, you can quickly and easily identify the correlations on which trials are most likely to close, which customers are at risk, where the most up-sell potential exists, and what are the best cross-sell opportunities – i.e., you can optimize the subscription lifecycle. Predictive analytics derived from the combination of usage and subscription data allows a company to know what actionable pricing and customer management opportunities exist and align their resources to optimize revenue growth in the Subscription Economy.