In some ways, we helped build the Customer Success category at SaaStr a generation or so ago. Some of the earliest writing here on Second Order Revenueand the power of a funded, dedicated CS organization informed a generation of B2B founders. We believed in it deeply.
So it’s not without some sadness that I say this: we haven’t had a genuinely great Customer Success experience in years.
Not worth taking the Zoom. Not even close.
What CS Has Become
The recent experiences tell the story better than any analysis does.
- A random CS rep we’d never heard from before emailed us that our pricing was increasing more than 10x (!). Thank you very much.
- A VP of CS sent us an NPS survey for a flawed product, got a low grade back, and replied with an automated response days later.
- A Director of CS we had never heard from before threatened to shut off access to our product early if we didn’t agree to a 20% higher renewal price.
And then there’s the other failure mode — the silent one. We had a vendor we’d paid $60,000 a year to for eight years. Last year we used them about half as much as the year before. No one ever reached out. Not once. Not a check-in, not a flag on a dashboard somewhere that turned into a human conversation. We didn’t renew. We didn’t even bother to complain. We just moved on. They probably still don’t know why.
In that same window, how many CS leaders reached out to genuinely ask how they could help us? When it wasn’t really a disguised sales pitch?
None. Zero.
The root cause in part: having CS report to Sales. This turns it into a sales function, not a success function. The rise of the CRO has accelerated this. When NRR is the only CS metric anyone tracks, the job quietly becomes “extract more money from existing accounts” with success language layered on top. The QBR is 45 minutes of upsell with a few slides about your health score. The NPS survey is a retention trigger, not a feedback mechanism.
That’s not Customer Success. That’s just Sales with an arguably friendlier title.
What’s Actually Working
Every single one of the more complex AI agents we run that is working well — and we run 20+ in production generating over $1M in revenue, spending $500K a year on agents versus $10K on Salesforce — had dedicated FDE time at launch.
Every one.
The FDEs at Salesforce, Replit, Monaco, Qualified, Artisan and others have been exceptional. And I mean genuinely exceptional, not politely-saying-something-nice exceptional. These are people who show up and actually solve the problem. They know the product cold. They write code alongside you. They debug on evenings. They figure out the integration nobody documented. They tell you when you’re doing it wrong and show you the right path.
That’s where the value is. Not the relationship layer. The technical layer.
Palantir invented the Forward Deployed Engineer model because they learned early that enterprise software deployments fail at the messy interface between the product and the customer’s real world. The demo works perfectly. The sandbox is clean. But your CRM has 80,000 duplicate records. Your Salesforce instance has 12 years of technical debt. Your “qualification process” is actually five different processes depending on which rep you ask. An FDE navigates all of that. A CS rep just asks you to fill out a survey about it.
The Hard Part: Can You Actually Afford This?
But … FDEs are expensive.
A good one costs $150K-$200K+ fully loaded, can only embed with a handful of accounts at a time, and the model doesn’t scale the way a CS team of 20 managing 200 accounts does.
So vendors are making a rational but suboptimal choice: ration FDEs to the largest accounts only. We’ve seen this directly. One major AI agent vendor we work with now only deploys FDEs to customers with 5,000 or more employees. Everyone below that threshold gets documentation and a Slack channel. Another has drawn the line at $100K ACV. Below that, you’re on your own.
I understand the math. You can’t put a $180K FDE on a $20K (or smaller) account, manage an agenetic deployment, and make the unit economics work.
But here’s what that math misses: the FDE isn’t a cost center. It’s your single best investment in retention and expansion. An FDE who gets a $20K customer to true production success — real workflows, real results, real ROI — turns that account into a $60K account within 18 months and a reference customer forever. The CS rep who sends a quarterly check-in email does not. Second Order Revenue is alive and well with FDEs, too.
The vendors limiting FDEs purely by company size or ACV are optimizing for the wrong variable. The right variable is deployment complexity and expansion potential, not headcount. A 200-person company going all-in on your AI agents needs an FDE more than a 10,000-person enterprise running a cautious pilot with three users.
At minimum, try not to do this. Find a way to get some FDE time to every new customer in the first 30 days. Even two weeks of real embedded technical help at launch changes the entire trajectory of the account. The cost is real. The return is higher.
This is the best investment in agent retention and success there is right now. Treat it like one.
What This Means If You’re Building CS
The old model to hire personable people to manage accounts, and escalate to technical when needed is broken for the current agentic environment. Customers are more technical. Deployments are more complex. AI products especially require real configuration and training, real integration work, real expertise to get value from.
A CS rep who can’t read an API response cannot actually help you anymore.
The vendors getting this right are blurring the line between CS and engineering. FDEs who own retention outcomes. Technical people compensated on renewal and expansion. CS leaders who can actually demo the product, not just present a slide about it.
The CS leaders who built their career on executive relationship management are going to find the next few years uncomfortable. That skill still matters. It’s just not sufficient anymore, and in many environments it’s not even the most important thing.
The FDEs are earning their keep. The QBR-deck crowd, increasingly, is not.
Is Classic Customer Success Dead?
To some extent, yes.
Slow-growth B2B leaders aren’t investing there. The headcount is being cut, the budgets are shrinking, and the function is quietly being absorbed into Sales or Support depending on which VP is willing to own the churn number. High-growth AI leaders, meanwhile, are investing in FDEs. Some, like Salesforce, are trying to repurpose CS teams into FDEs entirely. It’s a hard transition as the skill sets are genuinely different. But some CS people can make it, especially those who came up technical and drifted toward relationship work.
There will always be a support layer. Some tier of human contact for all customers, someone to answer the ticket, manage the renewal, handle the escalation. That’s not going away. But the golden days of CS as a strategic growth function, as the customer’s true ally inside the vendor, as a career path with real leverage and real respect — those are long behind us.
Long live its much more proactive cousin, the FDE.
There’s a certain irony worth noting. At SaaStr, we now have an AI agent we call QBee — our AI VP of Customer Success — that checks in on accounts every single day. Not quarterly. Every day. QBee does more proactive customer success work in a week than most CS teams do in a quarter. The best CS we’ve experienced in years isn’t coming from a human at a vendor. It’s an agent we built ourselves.
