The first part of a two part article on how to scale customer success, including a discussion on how to segment customers based upon customer outcome-related criteria rather than (or in addition to) supplier-related criteria



I was recently given the opportunity to talk to the delegates at the Customer Success Association’s 2019 European SuccessCon in London, and another of the speakers was Jason Noble, VP for Customer Success at Vinli and one of the two Jasons in the “The Jasons Take On…” podcast series. Jason is an influential thought leader in CS and also a great guy whom I count as a good friend within the CS community, and so I was really pleased to find he was giving a talk, and I am pleased to say that Jason didn’t disappoint me!

Jason’s talk was on the topic of customer segmentation – a topic which I generally try to avoid talking about on LinkedIn or elsewhere, since I rarely agree with any of the opinions expounded or ideas proposed, although I do deal with it at some length in my book Practical Customer Success Management. But Jason’s talk was different, because instead of talking about customer segmentation per se, he instead talked about a powerful alternative, and this alternative was (for once) something I can wholeheartedly get behind, because it is an approach which I also happen to advocate.

Jason’s basic premise is that the way to increase the range of situations in which Customer Success services can be deployed whilst remaining profitable is not to segment customers into the traditional “High Touch”, “Low Touch” and “Tech Touch” options, but rather to look at a blended or what he describes as a “Tech Assist” approach of deploying Customer Success Managers alongside supporting technology, where the blend is determined not by the size of the customer or the revenue value of the solution provided, but by the specific needs of each customer.

This concept provided inspiration for me, and the analogy of a DJ’s approach to mixing music that I will be describing in Part Two of this article came to me in a flash whilst Jason was talking – in fact I was even envisaging the knobs and faders on a mixing desk as he was speaking, so thanks Jason! – but we will get to “the DJ’s approach” in good time. Before that I want to set out the challenge and describe some of the common approaches to solving it and the potential pitfalls that might be involved with these approaches.

This article will be provided in two separate parts as follows:

In Part One (of two parts) I will set out the problem that CS teams face when attempting to scale customer success services to all or at least a larger proportion of the company’s customers. I will also discuss the traditional approach to solving this challenge, which is to segment customers by one or more criteria and then provide different CS services to each segment. I will then describe some of the issues or challenges that might come with this approach.

This will set us up nicely for Part Two, in which I will outline a different segmentation approach and describe how it works (and even provide a tool for doing so), and then I’ll go on to discuss three possible approaches to delivering CS services, which I call the AB Approach, the AB Plus Approach and the A to Z Approach, and I will finish by explaining how a CS team might determine which of these CS service scaling approaches would work best for them.


The Problem of Scale

I’ve wanted to talk about “the problem of scaling in CS” for some time, but the issue for me has always been that I struggle with the conversations around and descriptions of “Customer Segmentation” in the CS conversations and articles around this topic that I overhear or read about. The problem itself is in a sense a very straightforward one. It’s an issue of finite (and expensive) assets versus diminishing returns.

On the “assets” side of the problem we have the Customer Success Manager him or herself. The CSM is an expensive resource because they expect to be paid a salary for their work and because (contrary perhaps to what managers and customer stakeholders might like to think) they can only do one thing at a time. Compared with any other human being they’re about what you’d expect, but compared with automated software they are of course slow and expensive. On the other hand however, they are human beings (well, mostly) and as such they come with all the standard capabilities of relationship building and management, creative problem solving, innovative and intuitive thinking, presentation and influencing skills and so on that computer systems no matter how clever at raw data processing tend to struggle with, as well as other capabilities such as background research, data collection and analysis, report creation, process management, and so on that computer systems tend to be able to do very well. In other words for various reasons both emotional and practical, the CSM is very hard to completely replace by an automated system, particularly for larger, more complex and more customized or otherwise unique engagements where it would be difficult to “pre-program” a computer system to know what to do and/or where the development and ongoing management of interpersonal relationships are considered important or even essential.

On the “returns” side, we have the issue of the Pareto Principle to consider, which is often referred to as the “80/20 rule” which can be applied to many situations but in this case where 80% of all revenues (for instance) comes from 20% of all customer engagements (for example). In this situation, the relatively expensive but highly sophisticated and extremely versatile CSM can be deployed to help with the 20% of engagements that have relatively high value to ensure that this value gets attained. However, with the 80% of engagements that have relatively low value, there simply is not enough “meat on the bone” in terms of profits that can be diverted to paying for the expensive CSM’s time and efforts without the entire engagement ending up actually being an expense rather than making the company a profit.

So there we have it – how do we scale the undoubtedly (and hopefully provably) valuable but rather expensive customer success services to 100% of engagements rather than just the 20% of high value ones, so that the company gains the benefits of customer success activities that increase renewal rates, improve expand levels and enhance additional sales opportunities etc, yet without breaking the bank by throwing more and more high value yet high cost CSM assets at the problem?


The “Automation and Segmentation” Response

The standard response to this problem has been twofold. Firstly automation, and this we have already discussed and described in the previous section. This is where a software system is developed that takes over the role of customer success delivery from the CSM. The basic principle here of course is that unlike the issue of CSMs only having one pair of hands and therefore only being able to perform one task at a time, the software system can perform multiple tasks simultaneously – as many in fact as you are prepared to provide the necessary IT power for, such as memory, processing, network bandwidth, etc.. The beauty of the software system is that although it can be expensive to develop initially, once in place it becomes highly scalable and can deliver simultaneous CS services for as many customer engagements as necessary but without the commensurate increase in costs that would be associated with the same level of “scaling up” by employing more CSMs.

Of course compared with its human CSM counterpart, the CS software system is almost certainly going to be a lot more rigid and inflexible in its approach, less able to solve complex problems that require intuition and creativity, and a lot less able to develop deep and meaningful relationships with the customers’ key stakeholders (unless either your customers’ stakeholders or your CSMs are a little weird of course – something that perhaps we shouldn’t entirely rule out in all companies). On the other hand, the automated system might actually be better than the human CSM at some types of tasks, such as managing complex but standardized processes, and collecting, storing and analyzing raw data for example. Plus of course the software system is “always on” and is available 24x7x365, unlike most CSMs who generally prefer to be asleep at say 03:00 in the morning and often (I have heard) have the audacity to turn off their work phones to make them unreachable at this time,

The second part of the standard response has been segmentation, and this is where the CS team itself or perhaps more commonly a senior management team comprising for example CS managers, Marketing managers and Sales managers sit down together to review their company’s customers and to divide them up and group them into two or three separate sections or segments, where each segment contains customers that share a set of similar characteristics that the team defines as being important from a customer success services delivery perspective.

Once we have created these segments and placed our customers into them, we can now treat each segment differently. So for example maybe we create three segments called A, B, and C, and we place each of our customers into one or other of these three. Now what we can do is apply our different CS service types to each one. Perhaps in our example we provide a “High Touch” CS service for customers in Segment A, where they get a senior CSM either dedicated to just one customer or perhaps spread across a small handful of customers, providing a highly bespoke and feature rich CS experience for them. To continue our example maybe we provide a “Low Touch” CS service for customers in Segment B, where they get a limited amount of a more junior CSM’s time, and this CSM is perhaps spread across a much larger number of customers and as such can devote less time and effort to each individual customer when compared with the “High Touch” approach. This service therefore delivers a less bespoke and less feature rich and time intensive CS service than is provided for customers in Segment A, but nevertheless is still of value to the customers in that segment. To complete our example, perhaps we provide a “Tech Touch” CS service for customers in Segment C. With this service no CSM of any kind is provided at all, and instead these customers get the automated service provided by the software system we have described above, which can of course simultaneously deliver basic but hopefully still useful CS services to multiple thousands of customers at once.


The Issue of Segmentation Criteria

By combining segmentation with automation in the way illustrated, we can now solve the “How to scale our CS service?” conundrum. Our solution provides the “right” level of CS service appropriate to the needs of different customers, and delivers those CS services to all our customers, yet at the same time it does so without breaking the bank or eating too much into profitability by having to employ more CSMs than we can afford, and so everyone is happy. The customers are happy because they can pay a reasonable price for the product or service they need to purchase and get an appropriate CS service from their supplier. The supplier is happy because it gets happy and successful customers who renew, who expand their consumption, who make additional purchases and who act as advocates for its marketing activities and it can still make a reasonable profit on each deal.

Everyone is happy. What’s not to like?


Let’s say we go down this “automate and segment” route. What criteria might customers be segmented by? There is no simple answer to this question, and in fact its answer is often debated on CS-related forums and at CS meetups and training sessions. There are also lots of articles on this topic by CS experts with titles such as “How to Segment Your Customers to Maximize Revenues and Minimize Churn” that CS leaders can read to gain opinions from a variety of experienced and knowledgeable sources, and reading some of these is probably a good idea. These articles and the discussions and debates tend to come up with the following options for criteria to use when segmenting customers for CS purposes:

Recurring RevenueThe amount of recurring revenue currently earned from the customer from renewable contacts. This could be calculated monthly (MRR) or annually (ARR) and works well for SaaS companies and for other companies where a significant proportion of total revenues comes from recurring revenues.
Potential RevenueWhilst Recurring Revenue works well for companies with a large percentage of income coming from renewable contacts, it’s not so good for other types of businesses. Instead the total pipeline or predicted revenue (ie whether recurring or not recurring) that is anticipated to be gained from a customer over the next period (for example 12 months, or 36 months depending upon typical sales cycles) could be used.
Potential AdvocacyThis is where a customer either spontaneously and without being asked to do so provides great marketing by “spreading the word” – for example on social media. It also includes where specifically a customer is particularly useful as a reference site. This might be because of their brand name or reputation, or because they happen to provide a great reference case study for example.
Regional IssuesThis is a geographically based segmentation criterion, which straightforwardly divides up customers based upon where they are in the world. This might be useful when considering for example in-language requirements or cultural needs, or quite simply the geographical location of the suppliers’ CS assets compared with the customers’ stakeholders and the cost in time and money of the travel involved in getting them together.
Health ScoreThis is where the supplier determines to place more expensive resources (eg “real CSMs”) where they are most needed in order to prevent churn. A health score system ranks customers perhaps on a scale of 1 to 10 or using a simple Red / Amber / Green traffic light system to denote the level of risk of churn that each customer is determined to have, and more expensive resources can then be allocated to higher risk customers to try to win them back.


Of course there is nothing to say that only one criterion should be used. The segmentation team might decide to use multiple criteria, and this would certainly be a good idea in most situations in my opinion. However, I have come across very large companies (ie multi-billion dollar annual revenue sized MNCs) that segment based upon revenue alone. In this way, customers with large bank accounts and/or that place large orders get the “High Touch” CS service, customers with medium sized bank accounts and/or that place medium sized orders get the “Low Touch” CS service and small customers get left with the “Tech Touch” or “Automated” CS service. This approach is simple and straightforward to do, making it relatively clear as to what level of CS service to offer each customer, and helping the supplier to reach much more than just “the 20%” of the Pareto Principle by scaling CS operations to reach lower revenue value customers in an affordable way.


The Supplier Centricity Problem

There is a big problem with this in my opinion, and the problem is this – the approach is entirely supplier centric. What do I mean by this? What I mean is that the only consideration being taken here is the needs of the supplier, and the needs of the customer are simply ignored or at best assumed based upon size of company, amount being spent, location, and so on. In the customers’ eyes it might not unreasonably be perceived by them as a self-serving approach. For example, the customer that is of little to no value for a case study is treated differently to a customer that happens to make a great case study, yet they paid the same as each other for the product or service they purchased. Or the customer that is viewed as having less future revenue potential compared with the average is given a lower value or lower quality CS service than another customer that is viewed as having more future revenue potential compared with the average, even though again both companies paid the same money for the same product or service.

Does this strike you as being fair? The answer has to be “no”, but I concede that some people might make the perfectly valid point that “life is inherently unfair” and that’s just the way the cookie crumbles. Point taken, and I cannot but agree. And indeed when I originally described the potential issue I deliberately used the expression “self-serving” rather than the more emotive “selfish” for that very reason. After all, most companies are not charities and they need to do what they can to maximize profits for their owners, right? It is not unreasonable to expect a commercial organization to be self-serving, in fact arguably it has to be so in order to stay competitive. It may not be great for some customers to pay more for the same level of service (or perhaps the same amount but for a reduced service) compared to other customers, but it could be argued that this situation is still better for the customers than that of the supplier going out of business.

So it may not be (in fact to be blunt it isn’t) fair, but is it reasonable? Or perhaps even if reasonable, is it a good idea? Again, the point here is that by their nature commercial companies must act in their own best interests. However, alongside the short term value that might be gained by providing more favorable services to some customers and less favorable ones to others, the supplier should also think of long term value. Long term value is not about the specific “win this time” but about winning in the long run, and of course one aspect of this is corporate brand and reputation, and therefore it is also about the sentiment for the company from both existing and prospective customers, and indeed more widely from existing and prospective investors and employees too.

Being the type of company that customers want to do business with, that investors want to put their money into and that employees want to work for is very important. So the short term benefits of giving preferential treatment to some customers in return for favors (eg being used as a case study) needs to be carefully weighed alongside the long term benefits of growing and maintaining a reputation for doing what’s best for all customers, regardless of the profit or loss from the company’s own standpoint. In reality of course it needs to be a blended approach, where the company has a strong corporate ethic that informs its company policies and its corporate culture for “how we do things around here” and it also needs to have a flexible and creative attitude to spotting opportunities for things like a great reference case study from a particular customer and trading value to gain that opportunity, be that through a preferential service or a discount or some other way.

In addition, reality is generally much more complex than the model. The model states that if we segment our customers by one, two or three criteria, then all customers in Segment A have a specific set of similar needs to each other, and all customers in Segment B have a different set of specific similar needs to each other, and so on. However in my experience this is not always the case. Depending upon a fairly wide range of factors (and in particular perhaps on the types of products and services being sold) it may actually be the case that customers within Segment A or B are in fact found to vary widely in their customer success service needs. And if that is found to be the case then the whole process of supplier centric segmentation to determine a clever way to scale CS services to reach the maximum number of customers whilst retaining profitability comes crashing to the ground.


Coming Up in Part Two

I have rambled on for long enough – perhaps way too long – so there we must leave it for today.

In Part Two (which is already fully outlined and partially written, so hopefully it will not take me too long to complete) I will be responding to the challenges posed by the supplier centric approach by proposing and describing a different approach – a customer centric one. I will list the criteria that we might use for customer centric instead of supplier centric segmentation and show how this approach can lead to the same or better short term or tactical benefits yet at the same time avoiding the long term or strategic risks inherent with the more supplier centric route, and the potential for wide variation within a supplier centric segment described here in Part One.

Having described and explained this different segmentation approach, and having shared a model for how to score a customer based on the new segmentation criteria I am proposing, I will then go on to discuss three potential approaches to actually delivering CS services to customers at scale, which is where Jason’s “Tech Assist not Tech Touch” concept will enter in to the conversation.  I will describe each of these three approaches which I have labelled the AB Approach, the AB Plus Approach and the A to Z Approach and I will explain how they work and how a CS team or leadership team can determine which of these CS service scaling approaches might work best for them, given their own unique situation and requirements.