# Marcus Tello, VP of Customer Success at Fieldline (147-person B2B workflow SaaS) — read of customer-propensity-scorer-ai, May 23 2026

> 9 years in CS, currently managing 6 CSMs and a Gainsight instance that we've been meaning to replace for two years. I have a 44-minute commute each way and I listen to Effectively Wild every morning, which means I think about win probability models more than most people in my job.

## How I got here

Googled "customer health score without data team" after our head of RevOps told me for the third time she doesn't have bandwidth to build a propensity model in Looker. Found this page on the second results page. Not a paid ad, just an organic hit. Opened it during lunch.

## What I clicked first

The hero line landed: "Know which customers will expand and which will churn before they do." That's the job. That's the exact sentence I said in our last QBR. So I kept reading.

Then I hit "Real-time behavioral scoring" and two sentences later it says "compute propensity scores daily." Daily is not real-time. That's the first place I slowed down.

## Where I paused

The honest disclosure block. Specifically: "we don't have live customers on this idea yet." I read that three times. I came here looking for software to buy. This is not software. This is a strategy package for someone who wants to BUILD this software. The page never makes that clear in the hero. The entire top half reads like a product pitch for a CS tool. You get to the scoring section and realize you are not the end user they are selling to. The actual buyer is a founder who wants to start a propensity scoring company. I am the imagined customer of the imagined product, not the actual customer of this page.

## What I distrusted

A few things. "No data science team required" is in 90% of these pages. It means nothing without a walkthrough of what the actual setup looks like. Does "raw warehouse tables" mean I need an engineer to write a connector, or is there a UI? The integrations list (Salesforce, Segment, Mixpanel, Amplitude) sounds complete but it's also exactly the list every tool in this category puts on their homepage.

More than that: the builder rates pain intensity at 4/10 on their own product idea. If the people packaging this think the pain is a 4, why am I supposed to be excited about it? They published that number, which I respect, but it's doing real damage to the pitch.

And the year-1 Fermi estimate is negative $19,000. I understand what this section is trying to do. But a page that leads with "know which customers will expand" and then buries "you will probably lose money year one" in a scoring table is not building toward a close.

## What would convince me

If this were a real product I could subscribe to: a short screen recording of the Gainsight-or-equivalent showing how a CSM actually sees the score in their workflow. Not the integration architecture. The CSM's view, the alert they get, the account they click into. That's the proof I would forward to my team.

If this is what it actually is (an idea kit for founders): a single founder who used the dossier, built a version of this, and landed two paying CS teams. One sentence, one company name, a number. That's it.

## What I'd ask in an email reply

1. The page says "pulls from Salesforce, Segment, Mixpanel, Amplitude, or raw warehouse tables" -- when you say raw warehouse tables, are we talking about a no-code connector or does someone on my side need to write SQL and schedule a job?

2. Your scoring shows pain intensity at 4/10. You built this framework -- so you think the pain here is mild. What does a 7 or 8 look like in your framework, and why did you publish an idea you scored a 4?

3. Is anyone running this in production? Not "we're in talks with" -- is a CSM at a real company logging in today and seeing scores?

## Verdict: dismissive

Not because the idea is bad. The idea is fine, maybe good. But the page is selling two different things to two different people and does not know it. I came looking for a tool and found a franchise kit. The mismatch is too big to forgive in a cold read.

---
*Memo by skeptic persona, generated 2026-05-23. Studio breaks own self-grading loop.*