A lot of SMBs go from year to year without insights into the core drivers of ‘Customer Value’ beneath the typical revenue and margin P&L items. They have to oscillate between in-house hires and consultants to deal with shifts in the market in relation to digital marketing, technology, and media, but struggle to discern how to measure, manage, and analyze performance of these initiatives over time. In this post, we look at the fusion of Business Model Analytics with a Customer Value framework to solve this problem and structure the business model to outperform.
The word ‘value’ is probably the most overused and least quantifiable term used in everyday business lexicon. To deliver value is the core concept of each and every business model, whether for a Fortune 500 company like Apple or an individual service provider.
The problem is that most businesses do not have a simple and sustainable way to measure Customer Value beyond headline metrics such as revenue and margin. Thus, when looking at service providers or in-house hires to capitalize on new trends/shifts in the market, there is a structural gap that makes it difficult to decide whether to swerve right or left.
The term “Customer Value” pertains to a quantifiable and objective system to measure the monetary value within a customer base for non-contractual businesses, while Business Model Analytics looks at ways to structure a basic model for these purposes without needing to hire a team of Data Scientists. The result is better predictive insights coupled with a model to manage both expectations and resources to optimize the business model into the future.
Customer Value: Quantifying ‘Value’ with CBCV
The most valuable data set that a company has on its servers is related to any specific customer data. Looking at the post on Customer-Based Corporate Valuation (CBCV), we could see the incredible amount of both statistical analysis and data science that went into the creation of the models that ThetaCLV created.
The ‘predictive models’ – based on the concept of Predictive Analytics – were used to analyze public companies such as Farfetch, Slack, and others. ThetaCLV has developed these models around the concepts of ARPU (Average Revenue Per User), CAC (Customer Acquisition Cost), and LTV/CLV (Lifetime Value or Customer Lifetime Value). These particular models apply the concepts of LTV/CLV to non-contractual businesses (ie. not SAAS).
A lot of these concepts link to these business model attributes such as: word-of-mouth marketing (referrals), customer loyalty, gross margin, etc. It is quite common to look at these different attributes in isolation, especially when talking about ‘margin.’ Almost anyone in business who is worth their salt knows that good margins drives good business outcomes, and bad margins drive bad business outcomes, so doing anything to increase margins is generally effective at improving profitability. Except when you consider the effects on supply/demand mechanics related to referrals, loyalty, and word-of-mouth. These business model attributes need to be considered holistically, which is the general principle behind CBCV and ThetaCLV’s predictive models; these can be effective for both marketers and financiers.
There is a large body of research going back decades to show that loyalty – while generally considered a soft metric in business – is an excellent predictor of long-term growth in good economic times, and continuity (avoid bankruptcy, mass layoffs) in a downwards economic cycle as we are in now. These Loyalty Metrics can be systematically broken down into four different key items, which drive the model for measuring the ‘Customer Value’ across a certain Cohort of customers.
Insights gained from this type of model would be priceless for any small or medium business. The ‘would be’ qualifier is linked to the fact that it would be incredibly difficult for any business to develop this type of model on their own, and then ensure that the model can continue to update and improve on itself, which is the hallmark of machine learning.
The inherent complexity of managing the inputs (creating Cohorts of customers, delineating data points, structuring the data), defining the rules for the model (time series related to ARPU, CAC, Churn, and LTV), and massaging the outputs (determine most relevant analytics, analysis in relation to strategy) is too much even for many well-resourced businesses.
Data Science: Open-Source Customer Models
In the Business Modelling – Predictive Analytics post, we shared that there are models developed on Python by companies like Facebook (ie. Facebook Prophet) that can be used by anyone with the expertise to load the library onto either Python or R.
In the Facebook Prophet White Paper (Forecasting at Scale, non peer-reviewed, 2017), they publish the key tenets and assumptions that underpin their view on forecasting. In this case, it is a time-series based model that could be used as a forecast for businesses that have a high degree of seasonality and expectations of non-linear growth. Not all businesses with have a sales pattern that is ideal for Facebook Prophet, but this is just one example.
There is, however, a lot of model development in relation to Customer Value happening on open-source platforms that are accessible to anyone!
These types of open-source models would be extremely powerful for any company who had the ability to implement them. The point here is that there is a lot of open-source research and statistical models that are available for businesses to test. These models don’t have to be developed on their own, not does a business have to hire a firm to create a bespoke one.
Nevertheless, this type of data science work is still unobtainable for most businesses. As a starting point, we look at a middle ground in order to model better ways for businesses to capture more Customer Value and not waste money on resources that are not aligned to these models.
Measuring Customer Value – Identify the Trend, Structure BM Analytics
The middle-ground solution takes the concept of a predictive forecast and marries it to a few core Business Model Analytics that can be visualized holistically.
What any SMB can take from both this post and the post on Predictive Analytics is that there is an identifiable trend in sales data that needs to be visualized.
Even drawing out a sales graph like above by hand is a preferable way to start thinking about this concept. As any data series will have Trend – Seasonality – Noise patterns, the growth is almost always non-linear in some form or another for non-contractual businesses.
The visual forecast needs to be coupled with some specific Business Model Analytics. As there are so many possible ways to shape Business Model Analytics, it is recommended to pick the top 2 – 3 metrics that have the most relevance to the expected LTV/CLV into the future.
Example – a business is able to sketch out their sales forecast on an annual basis in a relatively simple way. Based on analysis of available Customer Value data, the business discovers a large cohort of repeat buyers that drive the top quartile (25%) of revenue growth. As it remained unknown to the business that this group of customers was driving the top end of their revenue growth, they were dedicating a lot of resources to New Customer acquisition and thus neglecting marketing activities that both converted New Customers into Returning Customers, and incentivized Returning Customers to make Referrals.
After distilling this detailed analysis into Business Model Analytics, it was determined that there were 3 Analytics/Metrics that the company would track on an ongoing basis to try and boost sales into the next year, adjusted for any seasonality and noise in the expected data.
In this instance, the Analytics are now in place to track the success of any given strategy.
When looking at the Business Model Canvas, there is the Key Activities section of the Canvas that defines the most relevant activities a business needs to perform in order to succeed. The question for the business in this example above will then become, do they have the resources/expertise in-house to execute around these Analytics?
For example, a shift in strategy towards driving Referrals and moving money away from Paid Advertising would be significant and require a new marketing and brand strategy. This would be an instance where paid consultants/contractors could carry the torch and help the business while the company goes through a transition period towards their new growth strategy.
The opposite could equally occur and a company will realize that while it has a large Repeat Customer base and high % of Referrals, the analysis above indicates that New Customers need to be attracted and that means a whole different set of Business Model Analytics. The strategy could call for a new in-house hire to ensure outbound marketing efforts towards New Customers are geared towards a mid-term outlook.
The goal is to create a simple model to track the expected outcomes vs. the actual outcomes without needing to hire a team of Data Scientists. While the methodology listed above may not resonate with everyone, at a minimum it facilitates a certain depth of thinking to go deeper into the drivers of Customer Value and how to to harvest that data into long-term outperformance.