When managing a reasonably large (500+) customer base, you have a bunch of issues to deal with when it comes to keeping your customers happy and paying their bills every billing period.
How to (Selfishly) Tier Out Your Subscription Customer Base
Among your customers, you could divide them up into the following categories with regard to their likelihood to pay you more, less, or the same over time:
- Who's doing so awesome they will pay you more money without your even asking
- Who's doing awesome and is likely to pay you more money if you actively upsell
- Who's doing awesome but is unlikely to pay you more money
- Who's doing well enough
- Who's doing OK
- Who's doing poorly
- Who's doing poorly but isn't likely to cancel
- Who's doing poorly and is likely to downgrade
- Who's doing poorly and is likely to stop paying you money
(* Some of this assumes an upgrade/downgrade path, or multiple products to cross-sell...not everyone has that. But everyone should eventually!)
How Can You Indentify A Customer In a Given Tier?
It's a hard problem. Some people would opt for statistical methods (e.g. regressions and clustering), some would go for usage-based scoring (e.g. feature x is worth 10 points, feature y is worth 50, anyone 50-100 is "doing OK"), and others would do more traditional customer-satisfaction surveys (e.g. Net Promoter Score). Which is right to tell where each customer should be slotted into the list above? Well, of course, it's a blended approach to generate a set of passively collected & generated metrics.
How can You Reduce the Error Rate of Your Success Metrics?
With any metrics-driven approach to measuring human activity, there's an error rate. Even if that error rate is small (5%, say) you're still leaving some customers under-served, and over-serving the wrong ones. It's not a good situation when your assessment of customer happiness is wrong.
To reduce error rate, a blend of passive, numerical/research-based methodologies plus a small amount of elbow grease on the account management side. That human element on the vendor's side of being able to understand a customer's situation and balance it with the passive/numerical metrics goes a long, long way. It can really lower your error rate, increase the accuracy of the tiering assessment, and ultimately make happier customers.
P.S. Why an onion? I think of the customer success tiers as slices of an onion, and you can peel it back to increase the accuracy of assessment.