Six people evaluate you for a senior design role, and they are not running the same evaluation six times. Each holds a different question. Each needs a different form of proof to answer it. You advance when your evidence answers all six, not when one of them likes you most.
For a candidate with a straight line — four years as design director at a comparable company — this is academic. The line reads without help. Yours needs translation at every joint, and each evaluator needs a different translation.
Belief Gates mapped the stages a candidate passes through in sequence. This piece maps the people sitting in the debrief at the end of it.
The coalition writes before it talks
Public evidence from Google's hiring research, Amazon's Bar Raiser process, and OPM structured-interview guidance shows at least four decision architectures operating at this level: consensus panels, independent approval committees, hiring-manager-plus-veto structures, and named executive decision owners.
One safeguard recurs across nearly all of them. Interviewers write independent assessments before the group discusses you. Disagreement gets preserved long enough to be inspected instead of talked away in the room.
That is what makes a nonlinear background expensive. Each evaluator writes from their own burden, so a gap in one assessment surfaces in the packet regardless of how strong the others are. The design peer writes about craft. The product partner writes about outcomes. The engineering partner writes about system reasoning. If your evidence closed one burden and left another open, the split is visible before anyone speaks.
In systems like Google's, the people making the final call may never have met you. They read the packet. An enthusiastic write-up from the interviewer who loved your forward-looking artifacts, sitting next to "no outcome evidence tied to her specific decisions" from the product partner, reads as a split. Splits at this level resolve conservatively.
The same mechanism preserves exceptional signal, which is the half that candidates underuse. A conventional director tends to generate adequate assessments across the board: meets bar everywhere, nobody advocating hard. Your background can produce lines in that packet that no design director generates. An engineering partner writing that you reasoned about agent traces and failure recovery at production depth. A compliance stakeholder writing that your Thermo Fisher work shows exception-first design in pharma at a specificity a SaaS design director cannot reach. Those assessments sit there attributed to people with distinct expertise. They do not get averaged into the hiring manager's general enthusiasm.
So every piece of evidence you present has to be restatable by someone who is not you. If the design peer cannot write "she demonstrated current craft judgment through [specific artifact] that showed [specific decision]," that evidence does not survive the debrief. If the product partner has to reconstruct your verbal explanation from memory to write anything at all, it dies in the packet.
For each coalition member you will face, state in one sentence what they should write in their assessment. If you cannot, the evidence is not yet in a form that survives.
Six burdens, six doubts
Recruiter or search partner
Needs to believe: you belong in the category — level, scope, role family — and can be explained to the hiring manager in one sentence.
A straight-line candidate makes this trivial. "Design Director at [peer company] for three years, twelve reports, reporting to VP Product." Categorization done.
Where you trigger it: "Head of Product" does not file into the design-leader drawer. BCG DV adds a second translation, since consulting titles often don't correspond to the same scope in-house. A 2026 CEPR study using recruiter clickstream data found candidates crossing into a different occupation were contacted 7% less often than equally qualified incumbents, with roughly 60% of the gap attributed to perceived skill mismatch.
Retire it: hand the recruiter the sentence. "Design director who led Alibaba.com's North America redesign at $50B+ GMV, most recently shipped AI-native enterprise tools as Head of Product at TinyFish." Level, scale, and the product title accounted for in one pass. Back it with a chronology where every transition carries a one-sentence reason and the scope at each stop is explicit. The recruiter does not need your arc. They need to file you and move you forward.
Hiring manager
Needs to believe: you can prevent the specific failure this role was created to avoid, and your prior results transfer to this mandate.
For the straight-line candidate, transfer is assumed; the last role looks like this one. Evaluation moves straight to chemistry, judgment quality, operating style.
Where you trigger it: transfer has to be argued. The hiring manager sees client work rather than product ownership, an enterprise platform at a different scale and context than a Series C AI company, and a product title that raises the question of whether you left design. Research on executive search found that functional diversity can signal adaptability while greater career mobility simultaneously raises competence and retention concerns.
Retire it: route your evidence by the role's unit of accountability — the thing that gets this person fired, covered in Issue 11. Function-building mandate: lead with Alibaba. You named the structural gap, built the mandate, ran three cross-functional sprints, +20% daily transactions. Ship-AI-product mandate: TinyFish context plus Agentic Labs as production proof. The hiring manager is buying implementability, so show the mechanism alongside the number.
Design peer or craft evaluator
Needs to believe: your design judgment is current, your craft bar is high, and you make other designers' decisions better.
Conventional portfolios contain recent shipped work with visible decisions. The peer can see what was chosen, what was killed, and how the candidate's judgment shows up in the artifact.
Where you trigger it: the Head of Product move, primarily. Practitioners who have gone from leadership back to craft describe confronting whether they still had it, whether the tools and practices moved while they were managing. Senior design postings at Atlassian and CZI list hands-on design judgment, prototyping, and craft feedback as explicit evaluation criteria. The peer is not testing whether you were once strong. They are testing now.
Retire it: current demonstrative work. Walk the forward-looking artifacts on screen and narrate rationale against the decisions, so that when the peer points at a state diagram and asks why, the artifact has already answered. Open an Agentic Labs system live. Take a whiteboard problem and let them watch the judgment run in real time. TinyFish gives you verbal context about the production constraints you navigated, but it is not an inspectable design case and cannot carry this burden.
The packet also works in your favor here. A conventional director shows competent recent work. Forward-looking artifacts operating at the frontier of AI interaction design give the peer something to write that they cannot write about anyone else on the slate.
Cross-functional partner (product or engineering)
Two partners, two different doubts.
The product evaluator suspects consulting metrics are engagement-level stories, numbers that describe the project rather than your contribution. Peter Merholz described the stereotype plainly: consultancy designers produce attractive concepts, hand the work over, and are absent for the thousand small decisions shipping requires. The engineering evaluator suspects your AI fluency is conceptual, that you can discuss agent architectures but have not reasoned through failure states, model behavior, and recovery at the system level. The distinction between AI-tool fluency and AI-product fluency lives here.
Retire it, product side: published cases that keep contribution, mechanism, and result separate. Alibaba: gap named, mandate built, three sprints led, +20% daily transactions. Thermo Fisher as Product Design Director at BCG DV: $20M+ margin, 100% partner adoption, exception-first design in pharma. They need to see that you owned an outcome, not that you were present for one.
Retire it, engineering side: system-level reasoning demonstrated live. State diagrams, failure recovery logic, agent traces and auditability discussed from TinyFish experience without needing to show the product. "She reasons at the system level about model behavior and failure states" is a line no design-only candidate produces.
Executive sponsor
Needs to believe: that they can defend the choice in a room you are not in.
The hiring manager evaluates whether you can do the job. The sponsor evaluates whether choosing you is defensible. Research on selection committees found that committees construct justificatory accounts for their preferred candidates, combining hard and soft qualifications into an organization-facing explanation of why this person fits the leadership category the company believes it needs.
The burden, stated the way the sponsor experiences it: can I repeat why we chose her when someone asks why we passed on the candidate with five years as a design director at a comparable company?
Straight-line candidates make this effortless. "Ran a team of fifteen at [peer company], shipped [recognizable product], references confirmed she's ready for more scope." Thirty seconds, unchallenged.
Your path does not compress into thirty seconds on its own. Consulting, enterprise, product title, AI pivot. Each transition is defensible individually, but the sponsor has to hold the whole sequence and reproduce it under questioning. The career mobility research cited above is hardest to counter when the person making the case is not you but a proxy repeating what they remember.
Retire it: write the sponsor's sentence for them. Not a paragraph. "She built trust architecture at Alibaba's scale, shipped production AI agents at TinyFish, and published the framework we're hiring someone to implement. No other candidate has done all three."
The sponsor does not need every transition. They need one comparative rationale that survives being repeated.
Confidence note: the mechanism is documented. Committees construct organization-facing justifications, independent approval systems require the case to survive past the hiring manager's advocacy, and external hires draw broader scrutiny than internal ones. But no public design-hiring source documents a specific candidate rejected solely because a sponsor could not articulate the rationale. Treat this as informed reconstruction of committee dynamics, not as a documented common event. The action is the same either way: make the sponsor's job easy, or lose to the candidate whose story needs no construction.
Clinical, compliance, or operations stakeholder (regulated tiers only)
Needs to believe: you understand irreversible decisions, human gates, auditability, and escalation as operational requirements with regulatory consequences, not as design principles.
Where you trigger it: "trust" used broadly, unattached to a specific regulatory or operational decision.
Retire it: Thermo Fisher as Product Design Director at BCG DV is your strongest regulated production precedent — pharma, six partners, exception-first design, 100% adoption. Red Cross is high-stakes disaster-relief operations including Service to the Armed Forces, not healthcare; say so rather than letting them assume. The Trust essay and the human-agent system design work add depth after the operational precedent lands, not before.
What gets you killed
"I bring a unique perspective from consulting, enterprise, and product." The sponsor cannot repeat it. It reads as positioning rather than capability. Replace with a transfer argument tied to this role's mandate.
Leading with breadth. The coalition does not average your experience; each member evaluates against their own burden. Breadth that fails to close any single burden leaves everyone partially satisfied and nobody convinced. Weak additions dilute strong evidence.
Describing TinyFish work in more detail when the room stays unconvinced. The design peer and product partner need something inspectable. Verbal description does not become inspectable through repetition. Use TinyFish for production context, technical credibility, and the bridge narrative, then stop.
Assuming the hiring manager's enthusiasm will carry the debrief. Where independent approval or veto authority exists, it will not. If the product partner wrote "no outcome evidence" and the engineering partner wrote "conceptual AI fluency only," strong intuition from one voice does not override the packet.
Using "trust" as a universal design principle rather than an operational claim. Compliance stakeholders and engineering partners hear the word differently than you use it. Attach it to auditability, reversibility, human gates, and escalation paths, or it registers as philosophy.
Telling a different version of your story to each evaluator. They compare notes. Inconsistency between what the design peer heard and what the product partner heard reads as unreliability, not tailoring. Keep the core narrative stable and vary the emphasis and the evidence selection.
Tier deviations
At AI-native companies (Anthropic, OpenAI, Suno), the engineering partner's burden often absorbs the product partner's. One person evaluates technical fluency and product reasoning together, and the bar on system-level thinking runs higher than elsewhere. When one person carries both burdens, layer system reasoning and outcome evidence in the same conversation rather than leading with one and hoping to reach the other. Moderate confidence: this comes from published interview processes and practitioner accounts, and these companies are revising their evaluation structures faster than any other tier.
At growth-stage platforms (Ramp, Gusto, Headway), the coalition may be three or four people rather than six, and the hiring manager holds more unilateral authority. The sponsor burden drops because the justification audience is narrower. The craft evaluator's burden rises, because there is less organizational infrastructure to absorb a weak design hire.
At enterprise platforms (Salesforce, Atlassian, Adobe), the coalition is largest and most formalized. Atlassian's published process shows differentiated sessions — portfolio review, Product Thinking, Craft Excellence, a Product-Engineering-Design triad, values — each with distinct evaluators. The sponsor burden is heaviest here because the justification audience includes several layers of leadership.
At healthcare and regulated companies (Ambience, Maven Clinic, Vanta), the clinical or compliance stakeholder carries genuine veto weight, and their burden is the hardest for a nonlinear candidate to anticipate because it runs on domain criteria that design experience alone does not address. Recognition cue: if the loop includes someone with a clinical, compliance, or regulatory title, or if the posting lists domain certifications and regulatory frameworks as requirements rather than preferences, the veto is real and that burden is not optional.
Two-directional red flags
The loop has no design peer. If craft is evaluated only by the hiring manager or a product partner, design judgment is not a real evaluation criterion, which means it is not a real organizational value either. The role will not give you the craft depth you are looking for.
Every interviewer asks the same questions. Undifferentiated evaluation means nobody briefed the coalition on distinct burdens. Expect a debrief where impressions get compared instead of evidence, and impressions favor the candidate whose path requires no explanation.
The sponsor is absent or unnamed. Ask who makes the final decision and who presents the recommendation to leadership. If nobody can answer, the sponsorship burden has not been assigned, and it will default to the hiring manager, who may not have the standing to defend a nonlinear hire.
You are asked to "just walk through your portfolio" with no guidance. An unstructured review lets each evaluator project their own burden onto the same material, and you lose control of which evidence answers which doubt. Ask what they most want to see before you start.
Quick-Take Card
Coalition posture: A senior design hire is decided by evaluators carrying independent burdens; satisfying one does not retire another's doubt.
Lead move: Build the sponsor's sentence for them — one comparative rationale they can repeat without losing fidelity.
Top landmine: Evidence that depends on your live explanation dies in the written packet. Make every claim restatable by someone who is not you.
Opening question: "Who will be in the debrief, and how does the final decision get made?" The answer tells you whether the coalition is structured or informal, and whether a veto exists.
Pattern-break cue: If every interviewer asks the same undifferentiated questions, nobody briefed the coalition on distinct burdens. Impressions will dominate, and impressions favor the conventional candidate.
- Atlassian's published interview structure: The only target company with a detailed public design-interview sequence, showing differentiated sessions for portfolio, product thinking, craft, and a cross-functional triad — worth studying as a template for how formalized coalitions actually distribute their burdens.
- Maven's split design mandates: Maven Clinic is currently hiring for both a consumer growth lead and a Senior Staff care-delivery designer with production-code and model-behavior requirements, which means the coalition composition and technical gates differ sharply even within the same company.
- Anthropic's eligibility boundary: The Evals & Prompts posting requires production-quality Python, evaluation pipelines, and regression suites — a qualification gate that no amount of narrative positioning can retire, and the clearest current example of where the burden-of-proof framework hits a hard stop.
- Career mobility's competing signals: A peer-reviewed study of 1,934 finalists across 378 senior searches found functional diversity can signal adaptability while overall mobility simultaneously lowers offer likelihood through retention and competence concerns — the empirical basis for why the sponsor's sentence matters more than the candidate's breadth.