When most people hear the word "pipeline," they think of sales. It makes sense; sales teams are trained to manage pipelines. They have the tools, the processes, and the Salesforce workflows, and their companies have invested in getting it right.
So why hasn't customer success caught up?
We've all heard that it's cheaper to keep a customer than to win a new one. Yet when it comes to building forecasting discipline in CS, the investment hasn't followed. Resources and training go to sales first, and CS teams are often left to figure it out on their own.
Customer success has a pipeline too. It's made up of different things: customer behaviors, business context, and signals that build over time and eventually become renewals, expansions, or churn.
The trouble is that most CS teams don't recognize that pipeline until they're already close to the outcome. By then, you're reacting to it, and it's too late to forecast.
Attention is the constraint
Pick one of your largest accounts. Every week, that account generates a huge amount of information:
- Logins and product usage
- Support tickets
- Meetings with your CSM
- Executive changes
- Budget shifts
- Hiring or layoffs
The list goes on. We have more customer data than ever before, and yet forecasting in CS is still unreliable.
That's because the forecasting limit is attention. No CSM can realistically investigate everything, and many haven't been trained to know what's worth investigating.
Too often, CSMs look in the rearview mirror when they should be looking through the windshield. They ask:
- Did they renew last year?
- What's the NPS score?
- What does the health score say?
Those questions have their place, but they only tell you what already happened. A forecast has to look ahead.
So, with everything available to you, what do you investigate?
The signal stack
To answer that question, I use a framework I call the signal stack. It sorts customer information into four types, and the order matters.
- Behavioral signals are what the customer is doing: usage, adoption, training completion, and engagement with your team. This is the most direct view of how a customer experiences your product.
- Contextual signals are what's happening around the customer: leadership changes, budget shifts, market conditions, and champion movement. These shape behavior, often before the behavior itself changes.
- Predictive signals are the models we build from the first two categories: health scores, risk scores, and renewal likelihood scores. We love to put a score on things in CS, and scores can be useful summaries. Just remember they're conclusions drawn from other signals.
- Historical signals are outcomes: renewals, expansions, churn events, CSAT, and NPS. They help with pattern recognition, but they come at the end of the story.
The sequence is intentional. Customer behavior happens first. Business context shapes that behavior. We build models from those signals, and then the outcomes follow. Once you see that order, it's much easier to see where your attention belongs.
