Customer Taxonomy

Dave Epperly

Hello team,

I'm interested to hear how others are classifying/sub-classifying their customers in order to uncover meaningful insights regarding segmentation, engagement best practices, and to inform ICP. 

As our business has grown in complexity the 'uniqueness' of our customer set has diversified. What began as a single customer 'type', within a single industry, using a single technology has evolved into an incredibly diverse customer set. My assumption is that there are meaningful distinctions buried within the noise. However, segmenting for everything is madness. 

So, who's dealt with this issue before? Where did this path take you and was it fruitful? How and where did you capture your data? Who captures the data and who maintains the fidelity of the data? Did you find your version of Darwin's Finches or did you just wind up keeping your S,M,L, buckets for the sake of simplicity?

Thank you for your insight,

Dave

Dan Balcauski

Hi Dave,

I deal with customer segmentation with clients all the time. It's a big topic (especially as you talk about data governance and infrastructure).

The very first question you need to ask is what decision are you trying to make. What would you do with this segmentation if you had it? Clarity in the decisions and business questions will help tremendously in prioritizing what needs to happen first. 

At a high level, the best way to segment your customer base is by behavior. In a software world, especially in SaaS, your best source of data is product telemetry systems. Once you have that data, augmenting it with all the "data exhaust" of customer data systems in the company (think support tickets, marketing automation systems, subscription management systems) can help further refine your profiles.

This can be quite an undertaking depending on how far along your company is on data infrastructure and practices. I would not diminish simple segmentation by size (e.g. ARR/license usage) as a starting point, as it often does reveal meaningful differences across those groups. 

In terms of infrastructure, that is a far ranging conversation. The first step is assessing what you have.  The next is figuring out if there is a way to get it centralized into a data warehouse. This will make it much easier for your BI analysts and/or data scientists to do their work.

Best,
Dan

Dave Epperly

Hi Dan, thank you for the thoughtful reply. 

We're fairly far down the segmentation and data gathering path- have great telemetry all along a well-mapped and traveled Customer Journey. 

Given this level of maturity and investment, we need to make the most of it. I probably could have been clearer when laying out the issue - but I'm essentially asking about approaching from the opposite direction:

Post segmentation, post ICP, post specific journeys (teams, etc.)... what happens if you take a single factor (Lifetime Customer Value for example), something you'd want to optimize for and work backward to the customer types that result in the highest likelihood to produce max LCV?

My guess is that you/we'd see specific customer types heretofore uncategorized that stand out - some combination of maturity in the space, use-case for our product, technology stack, and industry that we were formerly blind to (for example).

Having this data would be transformational. I think I have a path to get there but was just curious whether anyone else within the GGR community had traveled the same ground.

Thanks again,

Dave

Dan Balcauski

Hi Dave,

I see the confusion from the initial premise. What you're describing may involve customer segments, but doesn't necessarily. Ultimately, what you're describing has to do with customer insights. In the broadest sense, you can think of customer insights work in two domains: qualitative and quantitative. (We could also get into inductive/deductive, but don't want to turn this post into a novel). 

Qualitative
Available to every company, study success and failure (I find the best insights reside in the extremes). Phone/video calls give 95% of the value except in extreme cases where more ethnographic approaches are necessary. I highly recommend one-on-one - focus-groups/customer panels are a terrible way to get good data IMHO.

Quantitative
For high volume businesses (companies with >1000 customers), quantitative tools can be a great tool for exactly the problem you're outlining.  Leveraging the insights and hypotheses generated via your qualitative research, data models can be built against a target dependent outcome (like LTV) to figure out what factors drive success (or not) in your existing base. 

Obviously, this is just scratching the surface, but hopefully is more in the direction you were thinking.

Best,
Dan