I’ve been in three demo calls this month where prospects said “this looks great, but…” and then described a workflow that doesn’t exist in the product. Sales promised “we can customize that.” Engineering said “Q3 at the earliest.”

None of those deals closed.

The Pattern That’s Everywhere

Last week I talked to a Customer Success lead at a $50M ARR ERP company. They’re bleeding renewals. Not because the product is bad—because it’s inflexible.

A customer needs five approval stages. The product supports three. Engineering won’t prioritize it because only 15 customers need it. Those 15 customers represent $2M in ARR. They’re all at renewal risk.

The CS team knows exactly what needs to be built. They’ve documented the requirements. They’ve validated it with customers. They just can’t build it themselves.

So they file tickets that die in the backlog while customers churn.

This isn’t one company. I’m seeing this pattern everywhere in B2B SaaS right now.

The Death Spiral

Here’s how it plays out:

Your platform has an opinionated workflow. It works great if customers do things exactly your way. But no two enterprises run the same way.

Sales closes deals by promising customization. Implementation discovers it needs engineering. Engineering says “next quarter.” Next quarter becomes next year.

Meanwhile, the customer is working around your rigid workflows. Adoption stays low. At renewal, they say “we’re not getting value.”

You raise prices to compensate for churn. Higher prices with the same adoption problems accelerate churn. You’re in a death spiral.

Zylo’s data shows enterprises now manage 275 SaaS apps at $49M annual spend. Spending increased 9.3% while portfolios grew 2.2%.

Vendors raised prices 4x faster than companies added new tools.

When CFOs scrutinize that $49M portfolio, they’re asking “what’s our utilization rate?” If the answer is “60% of licenses are active but users only touch 30% of features,” you’re the first thing cut.

Where the Bottleneck Actually Is

The people who know what needs to be built aren’t the ones who can build it.

Your CS, Sales, Implementation, and Solutions teams understand customer workflows. They have relationships with stakeholders. They know what’s blocking adoption and expansion. They see the same patterns across multiple customers.

But they can’t deliver solutions. They can only file feature requests.

Customer needs a workflow adjustment? File a ticket. Customer wants custom approval routing? Add to backlog. Customer struggling because your UI doesn’t match their process? Send them “best practices” docs and watch the deal stall.

The people measured on retention, growth, and deal velocity are blocked from doing their job by engineering capacity constraints.

I watched this kill an expansion deal last month. A customer needed a custom dashboard to justify expanding from 100 to 500 seats. Implementation knew exactly what to build. Engineering timeline was 6 months. Customer stayed at 100 seats.

That’s $200K ARR lost because the team who could have closed it couldn’t build what was needed.

What Happens When You Remove the Bottleneck

I’ve seen a few companies figure this out. They gave customer-facing teams the ability to deliver workflow customizations directly—not through engineering, not after a roadmap process.

An Implementation team discovered 30 customers struggling with the same approval workflow. Instead of filing requests, they built a template using AI-assisted tools. Deployed it across all 30 customers in a week. Each customer customized it to their needs.

Adoption improved. Expansion conversations started.

A Sales Engineer was in a deal that needed custom routing logic. Instead of “we can build that in Q3,” they prototyped it during the POC using AI. Customer saw it matched their workflow. Deal closed. They reused the pattern for 50 similar deals.

A CS team identified renewal risk because workflows didn’t match evolved processes. They delivered adjustments in days. Customer renewed.

This is what happens when you remove the engineering bottleneck from customization.

The AI Piece

This wasn’t possible a year ago. Building workflow customizations required engineering knowledge, understanding data models, writing maintainable code.

AI changed that calculation.

Gartner predicted 70% of new apps would use low-code/no-code by 2025. They were right, but low-code still had a learning curve—proprietary tooling, data model complexity, maintenance overhead.

AI eliminates most of that. Instead of “learn our platform,” it’s “describe what you need.” Instead of “understand our schema,” the AI figures it out. Instead of “maintain custom code,” it refactors when things change.

Customer-facing teams can now build functional apps through AI assistance, governed by platform constraints. Not shadow IT—officially supported customization without engineering dependencies.

Even OpenAI’s chair acknowledged that “vibe coding” is a legitimate development approach now. The technology barrier is gone.

The Maintenance Objection

The obvious question: “Won’t this create unmaintainable sprawl?”

Which is valid. But consider what you have now.

Customers are already working around your rigid workflows—spreadsheets, manual processes, undocumented workarounds. That’s invisible technical debt you can’t track or improve.

With a proper system, customizations are visible, governed, and measurable. You can track what drives adoption, deprecate what doesn’t, scale what works.

Most workflow customizations follow predictable patterns anyway. Quote approvals. Deal reviews. Incident routing. The first customer’s custom workflow becomes a template for the next 50. You’re building a library, not bespoke software for each customer.

And AI dramatically lowers maintenance cost. Library deprecation, API changes, code updates—tasks that used to require specialized knowledge are now significantly easier. The “one person knows everything, then leaves” problem doesn’t apply when AI can explain the codebase to anyone.

Where This Doesn’t Work

Not every SaaS category needs this. If you have strong network effects, proprietary datasets, or genuine technical complexity (high-volume systems, serious compliance requirements), you’re probably fine.

But if your product is workflow software that sits on top of the customer’s own data? If the main value is “we built the UI so you don’t have to”? That value proposition is eroding fast when customers can build UIs that match their exact workflow in an afternoon.

The companies most at risk are ones where customer requests are consistently “can you make this work like X instead of Y?” If you’re constantly saying “that’s on the roadmap,” you’re vulnerable.

The Economics

McKinsey reports that B2B buyers increasingly demand performance guarantees and outcome-based pricing. They’re done paying for “powerful platforms” that sit mostly unused.

When prices rise 10-20% annually while budgets grow 2.8%, every dollar is scrutinized. Underutilized platforms get cut.

Your platform could drive massive ROI—if customers could adapt it to their workflows without waiting 8 months. But they can’t, so adoption stays low, and low adoption reads as “no ROI” when renewal comes.

The vendors who survive will be the ones who make it easy for customers to prove value quickly. Not in quarters—in weeks.

What’s Emerging

I’m watching a shift happen. Customer-facing teams are evolving from order-takers to value delivery engines.

ServiceNow built Creator Workflows. Salesforce opened Agentforce for partners to build custom agents. The pattern is clear: instead of fighting “we’ll build it ourselves,” they’re enabling it.

When a customer says “we can build this with AI now,” the winning response isn’t “but our product is better.” It’s “build it on our platform—we’ll give you the tools, security, and integrations. Build exactly what you need.”

You’re not competing with their AI-built tools. You’re the foundation they build on.

The companies getting this right have one Solutions Engineer building patterns that unlock revenue across hundreds of customers. One Implementation specialist solving adoption blockers that would otherwise take quarters to address.

Their competitors with rigid platforms and 6-month engineering cycles can’t compete. Their customers get told “we’ll add that to the roadmap.” These customers get working solutions before the next renewal cycle.

What Changes

When customer-facing teams can deliver customizations directly:

Deal velocity improves. Sales doesn’t lose deals because “we can’t do that workflow.”

Adoption increases. Implementation solves friction points in days instead of watching customers struggle.

Renewals defend. CS delivers adjustments that match evolved processes instead of watching utilization decline.

Expansion happens without headcount. Build once, deploy to 50 customers, each customized to their needs.

The technology exists today. The market pressure is real. Vendors raising prices while budgets shrink need to prove value faster than ever.

The question is whether you remove the engineering bottleneck before customers build their own workarounds—or before they build their own replacement.