The Courage to Build Coherent Products

Last week, I sat in on a roadmap review that felt less like strategy and more like arbitration.

Sales needed a feature to close a deal. Marketing wanted something flashy for the upcoming launch. A customer success manager had three urgent requests from three different enterprise accounts. And somewhere in the middle, a designer quietly asked, “How does this all fit together?”

No one had a bad idea. No one was being unreasonable. But the product was starting to feel like a group project where everyone wrote their paragraph in isolation.

At the same time, my feeds were full of familiar refrains: You don’t have a prioritization problem. You have a saying no problem. Case studies celebrating ambitious UX research initiatives. Essays dissecting incoherent social experiences. Product owners being asked when they’ll “add AI.”

Individually, these are practical conversations. Together, they point to something deeper.

We don’t just struggle to say no. We struggle to be coherent.

And coherence is a design problem.

The Hidden Cost of Always Saying Yes

In theory, prioritization is rational. Score impact. Estimate effort. Plot on a tidy 2x2 matrix. Ship the top-right quadrant.

In practice, it’s emotional.

Saying yes keeps the peace. It protects revenue. It makes stakeholders feel heard. It gives teams momentum. In early-stage startups, it can even be existential. As one founder recently put it, “At our stage, saying no can cost you the customer.”

That’s true.

But saying yes has a cost too—just a slower, more compounding one.

A 2023 Productboard survey found that 80% of product teams feel overwhelmed by feature requests, and more than half admit they ship items that don’t clearly tie to strategy. When I look at products that feel fragmented, I rarely see incompetence. I see accumulated yeses.

Each yes seems small:

One custom workflow for a high-value account
One experimental AI feature because "everyone’s asking about it"
One more filter, one more setting, one more toggle

Individually defensible. Collectively disorienting.

From a design systems perspective, you can feel it first in the interface. Patterns start to bend. Exceptions multiply. Edge cases become primary flows. Visual hierarchy grows noisy because everything is now "important."

But the deeper damage isn’t visual. It’s conceptual.

When a product tries to be everything to everyone, it stops being legible to anyone.

Coherence is what allows a user to form a mental model. Without it, they’re just memorizing buttons.

That’s not a prioritization issue. That’s a product identity issue.

Research Can’t Fix a Product That Won’t Choose

I’ve also noticed an uptick in beautifully documented UX case studies—capstones, portfolio pieces, deep dives into personas and journey maps. I love this work. Research is the backbone of responsible design.

But here’s the tension: research can surface needs. It cannot decide which needs define your product.

A few years ago, I worked with a team building a collaboration tool for distributed teams. Our research was robust—diary studies, usability tests, contextual inquiries. We uncovered real pain points:

Teams struggled with meeting overload
Documentation was fragmented
Onboarding new members was chaotic
Async communication lacked clarity

Each of these could justify a roadmap quarter. And in isolation, they all mattered.

The trap would have been trying to solve all of them equally.

Instead, we forced a harder conversation: What is the job we want to be exceptional at?

We chose onboarding clarity. Not because meetings or async didn’t matter, but because we saw disproportionate downstream impact. When onboarding improved, documentation improved. Meetings became more focused. Async threads made more sense.

This is where research and product judgment intersect. Research expands the landscape. Strategy narrows it.

The Anthropic report circulating this week—based on interviews with 81,000 people—highlights how differently roles react to AI integration. The more technical users push boundaries; others become cautious or skeptical. The insight isn’t just about AI. It’s about divergence.

Your user base is not one organism. It’s a collection of competing expectations.

If you try to satisfy all of them at once, you don’t get inclusivity. You get incoherence.

Incoherence Is a UX Smell

One of the Hacker News posts making the rounds was a tongue-in-cheek “guide to incoherent and isolating social experiences.” It read like satire. But it could just as easily describe half the digital products we use.

Incoherence in products tends to show up in four ways:

Language drift – The same concept is labeled differently across surfaces.
Interaction inconsistency – Similar actions behave in subtly different ways.
Competing hierarchies – Multiple “primary” actions fight for attention.
Phantom features – Capabilities exist but are buried, orphaned, or disconnected.

As designers, we often treat these as polish issues. They’re not. They’re governance issues.

When teams lack a shared definition of what the product is, every new initiative subtly rewrites the rules.

I once audited a SaaS platform that had grown through rapid enterprise customization. On paper, churn was low. Revenue was solid. But usability testing revealed something painful: new users consistently described the product as “confusing but powerful.”

That phrase should make any designer pause.

Power without clarity is intimidation.

We mapped the interface against its original value proposition. Over 40% of visible UI elements had no direct tie to the core use case. They existed to satisfy edge requests or legacy decisions. The design system had ballooned to accommodate them—new variants, special states, exceptions layered on exceptions.

Nothing was “wrong.” But everything felt heavy.

And heaviness is cumulative.

AI Is Amplifying the Coherence Test

The pressure to "add AI" is making this dynamic sharper.

AI features are genuinely different. They’re probabilistic. They’re opaque. They introduce behavioral variability into systems that used to be deterministic.

When you bolt AI onto a product without re-examining its core promise, two things happen:

Users lose predictability.
Teams lose clarity about what success looks like.

In a recent internal review, we evaluated an AI-assisted workflow that generated recommendations inside a structured tool. The feature demoed beautifully. But in usability sessions, users hesitated.

Not because the outputs were bad. Because they didn’t understand the relationship between their input and the system’s suggestion.

The AI wasn’t incoherent on its own. It was incoherent within the existing mental model.

Adding intelligence to a product that lacks conceptual clarity is like increasing the volume on a song that’s already out of tune.

If anything, AI raises the bar for coherence. When behavior becomes less predictable, your framing, language, and interaction patterns must become more deliberate.

Otherwise, users experience the product as moody rather than helpful.

Designing for Coherence (Even When It’s Hard)

So what does it actually look like to design for coherence in a world that constantly pushes you to expand?

From my experience, it requires four ongoing practices.

1. Define the Non-Negotiable Core

Articulate—in plain language—the primary transformation your product exists to create.

Not a list of features. Not a vision statement. A sentence.

When a request comes in, ask: Does this strengthen, extend, or dilute that transformation?

If you can’t answer clearly, you’re already drifting.

2. Treat the Design System as a Strategic Artifact

A design system is not a component library. It’s a codified expression of what your product believes about interaction.

When exceptions multiply, pause. Each one is a signal that strategy and execution may be misaligned.

Consistency isn’t aesthetic rigidity. It’s cognitive kindness.

3. Create Space for Explicit Tradeoffs

In roadmap conversations, I’ve started asking a simple question:

"If we say yes to this, what are we implicitly saying no to?"

Make the cost visible. Not just in story points, but in conceptual clarity.

Sometimes the right decision is still yes. But it’s an informed yes, not a reflexive one.

4. Measure Legibility, Not Just Adoption

We track engagement, retention, conversion. All critical.

But we rarely measure whether users can accurately explain what our product does and why it works the way it does.

In moderated sessions, I often ask participants to "teach the product" back to me. If their explanation is scattered or overly tactical, it’s usually a sign that we’ve optimized tasks without clarifying the whole.

A product should feel understandable before it feels impressive.

The Emotional Side of Coherence

Here’s the part we don’t talk about enough: coherence requires courage.

It means disappointing someone. It means risking short-term revenue for long-term clarity. It means trusting that depth beats breadth.

It also means accepting that not every user is your user.

That’s uncomfortable. Especially for teams that care deeply. Especially for designers trained to empathize widely.

But empathy without boundaries becomes chaos.

The most coherent products I’ve worked on weren’t the ones with the biggest roadmaps. They were the ones where the team could finish each other’s sentences about what the product was for.

That alignment shows up everywhere—cleaner interfaces, clearer copy, faster onboarding, more confident decision-making.

And users feel it.

They may not articulate it as "conceptual coherence." They’ll say things like, “It just makes sense.” Or, “I didn’t have to think too hard.” Or my favorite, “This feels built for us.”

That feeling is not accidental.

It’s the result of a thousand small no’s in service of a larger yes.

As our tools grow more powerful—AI-infused, highly customizable, endlessly extensible—the temptation to expand will only increase.

The real discipline of modern product design isn’t just prioritization frameworks or research rigor.

It’s the courage to choose a shape—and protect it.

Because in the end, users don’t experience our roadmaps.

They experience the story our product tells.

And that story only holds together if we do.