The Counter-Interview — What to Diagnose While They Evaluate You

The default failure mode at this level: treat the interview as a one-directional performance until the offer arrives, then scramble to assess fit during the negotiation window. Leverage is highest then. Information is lowest. Reverse that sequence. Start diagnosing in the first conversation. By the time an offer lands, the decision should take an hour.

Four dimensions below. Each has questions you can plant without sounding like an auditor, recognition cues that surface from how interviewers talk and behave in real time, and a kill criterion. The kill criterion is the point.

Whether They Actually Want a Leader Who Builds

Your site frames this plainly: strategic scope, hands-on craft. You lead design functions and you stay close to the work. Many companies claim they want that combination. Watch what they respond to when they see it.

Your portfolio is the primary diagnostic instrument. Three published pillars give you three axes to watch. The Agentic Work track, including Carrier IQ, Retail Velocity, and Brand Pulse plus the Trust Is the New Interface essay, proves you build with your hands. The Alibaba case proves you lead enterprise-scale design organizations. The 0-to-1 cases, Thermo Fisher through Allē, prove you ship high-stakes products from nothing to live. Which pillar the interviewer gravitates toward tells you what they're actually hiring. Which pillar they ignore tells you what they don't value.

If nobody across the loop engages the agentic work, they want a manager. If nobody asks about the Alibaba org-leadership layer, they want an IC they're calling a director. Both findings are useful. Both are better to know now.

Questions to plant:

• "Which of these tracks is more relevant to what you're solving right now?" Plant this in the portfolio review. It forces the interviewer to reveal whether they want craft contribution or organizational management. If they point to the agentic work and ask about interaction patterns, agent trust mechanisms, the Watch-Verify-Delegate progression, they want a leader who builds. If they point to the case studies and ask only about team size and stakeholder management, they want a manager whose portfolio checks a box.
• "When your current design leaders prototype or make direct design decisions, how does the team respond?" Save for the hiring manager or a peer designer. Present tense with specifics means it happens. Future tense or "we'd love for someone to" means it doesn't.
• "What's the last design decision that was made by a leader rather than delegated?" If nobody can name one, the role is managerial regardless of what the posting said.

Recognition cues:

The portfolio review tells you everything if you read the right layer. When the interviewer engages the Alibaba case by asking about trust-signal architecture across 200K+ suppliers or the agentic sourcing loop, they are reading your craft decisions. When they ask only "how big was the team" and "who were your stakeholders," they are reading your org chart. Both legitimate questions. But if nobody across the entire loop asks a single craft question about the work itself, high confidence the role will not let you do craft work.

Subtler cue: the gap between how the design evaluator and the cross-functional evaluator engage your portfolio. There's anecdotal evidence from Ramp loops where a design lead asked sharper component-level questions than the PM interviewer. That gap is diagnostic. When design and product are far apart on what this role does, the company hasn't resolved what it actually wants. You'll spend your first six months being the resolution.

Tier-level pattern: AI-native companies (high confidence) value the hybrid and may require it, but "craft" in those rooms often means prototyping with models, not visual systems. Enterprise companies (moderate confidence) say they want the hybrid but structurally reward the management side. Growth-stage companies are the wildcard: the hybrid is genuinely needed but may not survive the next reorg.

Kill criterion: Every interviewer treats the portfolio review as a credential check. No one asks about a design decision, a tradeoff, or a system-level choice across any of the three pillars. The role is purely managerial. Your hands-on contribution will be tolerated at best.

Whether a Design Culture Exists Beyond the Org Chart

You want a living design culture you can participate in and shape. Not a team that exists on paper while individuals sit isolated in product pods with no connective tissue.

Questions to plant:

• "How do designers here see each other's work?" Plant in the recruiter screen. Low-stakes, easy to ask early. The recruiter's comfort or discomfort with the question is itself a signal. The answer reveals whether critique, review rituals, or shared standards exist. If the answer is "we have a Slack channel," that is your answer.
• "When was the last time a design decision in one product area changed how another team approached their problem?" This tests for cross-pollination. Atomized design orgs cannot answer it.
• "Who on the design team would I be learning from?" Forces the interviewer to name specific designers and describe what makes them strong. If they can't, the design community is thin or invisible to leadership.

Recognition cues:

The most reliable in-conversation signal: whether designers appear in your interview loop at all. Adobe's design leadership explicitly recommends candidates ask who will be present and what their roles are. Do that at every company. Normal, expected, diagnostic.

Portfolio reviews are consistently present across company types, but whether a peer designer evaluates that portfolio varies widely. Some loops are entirely cross-functional: PM, engineering, hiring manager. If no designer evaluates your work during the entire process, moderate confidence that design as a community function is weak.

When interviewers describe collaboration, listen to the prepositions. "Working with engineering" is structurally different from "supporting engineering." The first implies parity. The second implies service. People use these words without thinking about them. That's why they're reliable.

Tier-level pattern: Large enterprise companies (high confidence) tend to have established design communities with critique rituals, design systems teams, cross-team review practices. The risk there is calcification, not absence. Growth-stage companies (moderate confidence) often have strong informal design culture that hasn't been formalized. AI-native companies are the most variable: some have deeply embedded design practices, others have no design connective tissue at all because the function is too new.

Kill criterion: No designer appears anywhere in the loop. When you ask about critique or review practices, the answer is vague or aspirational. Design is atomized. You will spend your first year building connective tissue the organization may not want built.

Whether the Domain Is the Point or the Backdrop

Your strongest published work lives in specific, high-stakes vertical domains. Pharma supply chain where a missed exception costs 3-5x more at the delivery gate. Crisis response where downtime means delayed disaster relief. Procurement at scale where payment trust and regulatory variance across markets are the design problem. You want that specificity. A horizontal platform role optimizing generic workflows across interchangeable use cases will underuse you.

Questions to plant:

• "What's the worst thing that happens when this product fails for a user?" Real consequences to real people in a named domain means you're in a vertical. Churn metrics or NPS means you're in a horizontal.
• "How much domain knowledge did your last design hire need to acquire, and how long did it take?" Save for the hiring manager. Companies that care about domain depth have opinions about onboarding timelines. Companies that don't will say "you'll pick it up."
• "Who are the three user types I'd be designing for, and what's different about their workflows?" Vertical domains produce specific, named user populations with specific constraints. Horizontal platforms produce personas.

Recognition cues:

Listen for constraint language versus flexibility language. Vertical-domain interviewers talk about what they can't do: regulatory limits, compliance requirements, safety thresholds, clinical protocols. Horizontal-platform interviewers talk about what they could do: extensibility, configurability, platform plays. Both valid businesses. Only one is where your portfolio proof is strongest.

The first five minutes tell you. If interviewers name a specific industry, specific user population, and specific stakes without you asking, the domain is central to the company's identity. If they describe the product in abstract platform terms and the domain only surfaces when you probe, the domain is incidental.

The third cue requires you to act. When you reference the Thermo Fisher pharma supply-chain context or the Alibaba regulatory-variance problem, watch the interviewer's response. Do they lean in with domain-specific follow-ups? Or do they generalize it back to "complex workflows" or "enterprise scale"? If every interviewer abstracts your domain work into generic complexity language, high confidence the daily problems will be generic too.

Tier-level pattern: Healthcare and vertical SaaS companies (high confidence) treat domain depth as a first-gate filter. They want constraint language in the first three minutes. Enterprise platform companies (high confidence) do the opposite: deep vertical attachment reads as narrowness. AI-native companies split: those building for regulated verticals care about domain; those building general-purpose tools do not.

Kill criterion: Interviewers consistently describe the product in horizontal-platform language. User populations are generic. When you ask about domain-specific constraints, the answer is that they're "still figuring that out." This is a platform design role wearing a vertical label.

Whether the Company Moves at the Speed It Claims

You shipped three products in three months at TinyFish. That velocity is the benchmark.

The interview process itself is the first and best diagnostic. Before you ask a single question about operating speed, their operating speed is already on display.

Recognition cues from the process itself:

Scheduling pace between rounds. Interviewer preparation quality: have they looked at your site, or are they scanning it during the call? Decision speed after each round: did the recruiter follow up in 24 hours or five days? Count the rounds that exist to gather signal versus rounds that exist to satisfy a stakeholder who needs to feel included. The interview process is the company performing at its best. It only gets slower from here.

Calibration benchmarks:

Company	Speed Signal	Observed Process
Ramp	"Build without permission"	4–5 design rounds, ~19 days avg (range varies widely)
Maven	"Move fast" as a value	~30 days application to offer
Anthropic	—	3–5 weeks for design loops
Front	"Speed" as core value	Posting asks for "structure without bureaucracy"

Use these as baselines. A company that claims Ramp-like velocity but runs an Anthropic-length loop is telling you something about the gap between aspiration and operation.

Questions to plant:

• "What shipped last month that wasn't planned three months ago?" Fast companies answer immediately with a specific example. Slow companies describe their planning process.
• "When a design direction is set, how many people need to agree before it moves to engineering?" Plant with a peer or hiring manager. Count the names. Three or fewer is fast. Five or more is committee governance.
• "What's the longest a design project has been stuck waiting for a decision that wasn't yours to make?" Save for the hiring manager. Requires enough trust to get an honest answer. What you hear reveals the actual bottleneck structure.

Tier-level pattern: Pre-IPO growth-stage companies (high confidence) run faster loops and faster operating cadences, but speed without resolved design authority means you ship fast on someone else's terms. Large enterprise companies (high confidence) move slower, and the slowness is structural, not fixable by one hire. AI-native companies are fast but volatile: the speed is real, and so is the likelihood that priorities reverse in a week.

Kill criterion: The process itself takes more than five weeks with repeated scheduling delays, interviewers who haven't looked at your portfolio before the conversation, or rounds that exist to satisfy a checklist rather than gather signal. High confidence that daily operating speed will be slower than what you need.

Running All Four

These dimensions interact. The interactions matter as much as the individual scores.

Strong speed with weak design community means you'll ship fast with no critique, no creative challenge, no one who pushes back on your work. Craft plateaus while output accelerates.

Strong design culture with weak domain depth gives you creative peers who sharpen your thinking on problems that don't carry real stakes. The sharpening is real. It just never translates to the kind of work your portfolio proves you're best at.

Strong domain depth with weak hybrid acceptance puts you close to consequential problems but prohibits you from touching the design work that solves them.

You don't need dedicated time for each dimension. The portfolio review covers Dimension 1 naturally. A single question about how designers see each other's work covers Dimension 2 in ninety seconds. Domain language surfaces in the first five minutes without prompting. Dimension 4 data accumulates passively from the moment the recruiter responds to your first message.

After each conversation, document a simple grid: four dimensions, green/yellow/red. By the second round, the pattern is usually clear. By the final round, if any dimension is red, you know.

The hardest discipline is acting on a kill signal from a company you're otherwise excited about. Do it anyway. I have watched too many senior design leaders take roles where the daily reality didn't match the pitch, then spend two years fighting for something the organization never intended to give them. These four dimensions exist to prevent that. Trust the signal. Even when the title is right, the comp is right, and the company name would look good on your LinkedIn.