Start here
The Allē case has the strongest strategic premise in the portfolio and the weakest claim-evidence chain. The relationship argument works: transactional points system to coordinated loyalty architecture, consumer and provider unified under one design logic. Seven design decisions, real craft, extensive prototyping across both surfaces.
The problem: three of your four headline metrics float above the case without a causal thread connecting them to anything you did. The fourth ($42 CAC) holds because it names a decision, a mechanism, and a before-and-after figure in the same section. The other three are exposed the moment a panel asks "how do you know your design caused this?" At Director+, that question comes up in most debriefs.
The most urgent item is the member count. AbbVie's own press releases report eight million Allē members as of April 2026. Your case claims 30 million. The case doesn't define what that number counts: current members, cumulative enrollments, accounts inherited from Brilliant Distinctions, something else. A hiring panel that searches "Allē members" will find AbbVie saying eight million. Your case says 30 million. That's not a rounding difference. Resolve this before the case goes in front of anyone.
Lens 1: Single strategic premise
What works. The opening line sets it up: "30 million enrolled members. A loyalty program almost none understood." Followed by a first-person attribution statement: "I redesigned both surfaces and defined the strategy connecting them." The body sustains one argument throughout: each decision section names a prior state and a redesigned state, and the transformations share a logic (points-balance to progress, catalog to care plan, appointment to evidence, billing tool to patient context). This is a coherent design argument.
What's missing. The premise promises a system. The evidence delivers sequential decisions. The teaser claims "one information model, not two products," but neither the teaser nor the full case shows that information model. There's no diagram, data flow, or end-to-end transaction showing a consumer action triggering a provider state change. The shared concepts (tier status, expiration, treatment history) appear in both tracks, but the case doesn't demonstrate them as the same data surfaced differently. It demonstrates parallel design decisions that use similar vocabulary.
Revision. Add one view showing how a consumer's expiring points appear simultaneously in the consumer's progress screen and the provider's urgency-sorted patient list, driven by the same underlying event. You have the prototypes. The connection just isn't drawn.
Lens 2: Decision rationale depth
What works. Every numbered section explains the problem-to-choice move. That's better than most portfolio cases at this level.
What's missing. The reasoning stops at "we chose this because it was better." Across seven decisions, the case documents zero alternatives considered and rejected, zero research inputs (sample, method, finding), zero constraints that forced a tradeoff.
Compare this to your own work. Alibaba runs 32 interviews into three named design bets with sprint-level results. Equinox+ documents 75 hours of research across four cities, names the prioritization logic, and describes a late engineering-capacity tradeoff that changed the shipped product. Allē has more craft artifacts than either of those cases and less decision-making evidence.
The visual-system explorations are the clearest example. Eight directions (Noir, Blush, Prism, Veranda on consumer; Analytics, Daylight, Ember, Canvas on provider), each demonstrated through three-state prototypes. No record of which was selected, what evidence compared them, or what tradeoff determined the outcome.
Revision. Pick two decisions. For each, add one sentence naming the alternative that lost and one sentence explaining why. The visual-system explorations are the easiest place to start — eight directions shown, none selected. Even a line like "Prism scored highest on accessibility but Blush better matched provider expectations around clinical credibility" would convert those explorations from a gallery into a decision record. You don't need to reconstruct a full research narrative. You need to show that the choice was a choice.
Lens 3: IC + manager proof
What works. The hero identifies you as Product Design Director leading three product designers, UX research, and brand. The body demonstrates extensive hands-on design work at the component level across two surfaces.
What's missing. The body doesn't demonstrate a single act of leadership. Across the entire case, there's no mention of hiring, coaching, critique protocols, delegation rationale, quality standards you set, or executive alignment you drove. The title says director. The evidence reads senior IC.
I've written about this oscillation problem before. When two interviewers write down two different candidates — one sees a director, one sees an IC — the debrief resolves the discomfort by advancing the simpler candidate everyone can describe in one sentence. You become harder to describe.
Equinox+ handles this with an explicit operating-scope block that separates team composition, partner set, owned decisions, and influenced decisions. Allē needs something equivalent.
Revision. Add a concrete paragraph: what you directed, what you delegated, what your three designers owned, what you escalated to leadership. A scope statement, not a management essay.
Lens 4: Consumer and provider as unified logic
What works. The case presents consumer and provider inside one chapter structure. Shared concepts (tier status, expiration, treatment history, redemption) appear in both tracks. The premise frames them as one system with two expressions. This is the right argument.
What's missing. The evidence doesn't match the argument. Consumer decisions appear as Track A, provider decisions as Track B, presented sequentially. The information model that would prove they're one system is described but never shown. A reader can see that both surfaces use "expiration" as a concept. A reader cannot see that they share a data layer, that a consumer event produces a provider-side state change, or that the design decisions on one surface were constrained by decisions on the other.
Revision. The artifact is the same one Lens 1 calls for: a diagram or annotated flow showing a single loyalty event across both surfaces. What it uniquely proves here is that a design decision on one surface was shaped by a condition on the other — that the provider's urgency sort, for example, couldn't be designed without knowing how the consumer's expiration notification would change behavior. Without that, dual-surface describes your scope, not your design logic.
Lens 5: Outcome attribution
This is where the case is most exposed. Four headline metrics. Here's what the body actually supports for each:
| Metric | Status | Action |
|---|---|---|
| $42 CAC | Supported | Keep as-is |
| 47% reactivation | Contradicted by body | Fix label or fix hero |
| 3.2× redemption | Unsupported | Connect to a decision or demote |
| 30M members | Unsupported, externally contradicted | Define or replace |
$42 CAC. Supported. The care-plan section names the decision (in-app treatment planning), the mechanism (intent-first conversion replacing front-desk enrollment), the comparison (2.4× the conversion rate), and the result ($92 to $42). Missing: measurement period, denominator, channel mix. Reasonable asks in a live interview, not fatal gaps on the page. Keep as-is.
47% reactivation. Contradicted by its own body. The hero labels this "lapsed reactivation." The detailed section labels 47% as the open rate for expiration push notifications. The section separately reports that 68% of reactivating members named the notification as their reason for returning. These are three different numbers measuring three different things: reactivation rate, notification open rate, self-reported attribution. The hero and the body can't both say 47% and mean different things. Fix the label or fix the hero.
3.2× redemption. Unsupported. This figure appears in the hero and metadata. It does not appear in any decision section. The body never defines "redemption," provides a baseline, names a comparison period, or identifies which design decision drove the lift. Your teaser adds "+0.4 vs prior" beside it, detail that doesn't exist in the full case. Connect it to a specific decision with a mechanism, or demote it from the hero.
30M members. Unsupported and externally contradicted. The body mentions 18 million enrolled members at the 2021 relaunch but never explains how 18M became 30M or what design decision contributed to that growth. AbbVie's press releases from January 2024 through April 2026 report between six and eight million members. If your 30M figure uses a different definition (cumulative enrollments including Brilliant Distinctions, total accounts ever created, an internal metric that counts differently than AbbVie's public reporting), the case needs to say so explicitly. If it can't be reconciled, it needs to come out.
I flagged this conflict in the standing positioning model as unreconciled. It remains unreconciled.
Revision. The custody rule applies: one claim, one evidence carrier. For each metric you keep in the hero, the body must contain a section where you can trace design decision → mechanism → measurable result. $42 CAC passes. The other three currently don't. Fix 47% by aligning the label. Fix 3.2× by connecting it to a decision or removing it. Fix 30M by defining it or replacing it.
Lens 6: Healthcare-adjacent framing
What works. Nothing. The case doesn't engage this dimension at all.
What's missing. The interfaces show BOTOX, Juvéderm, injection units, treatment photographs, follow-up timing, aftercare information, and a "Call doctor's office" action. The page describes treatment-timing prompts as "scientifically specific" and the progress record as "clinically useful." But the case contains no discussion of patient consent for photographs, health-data privacy, adverse-event escalation, the boundary between loyalty messaging and medical advice, or pharmaceutical promotion constraints.
AbbVie's own Allē documentation includes consent conditions, prescription-product safety information, and financing disclosures. Their medical-aesthetics communications include boxed warnings. These constraints shaped the design environment you worked in. The case acts as if they didn't exist.
This matters across archetypes, but it's a threshold failure for healthcare and regulated-industry roles. Those panels test consequence literacy: what breaks, who absorbs it, what guardrails you built. A loyalty program that handles treatment photographs, injection records, and provider-patient communication under a pharmaceutical parent company is a rich environment for demonstrating that literacy. You're leaving it on the table.
Revision. Add one paragraph documenting one real constraint from the regulated environment and how it shaped a design decision. Consent for treatment photos, the line between loyalty notification and medical advice, HIPAA considerations in the provider campaign builder. Pick whichever you can describe most concretely.
Structural notes
The approach heading promises "Two tracks. Six design decisions." The summary matrices list eight transformations (four consumer, four provider). The body contains seven labeled sections (three consumer, four provider). Pick a number and make the page match it.
The hero dates the work to 2020. The body says the relaunch happened in 2021. The teaser says 2021. Align them.
Three highest-impact revisions, ranked
1. Resolve the 30M member figure. If you can define what it counts and why it differs from AbbVie's public eight-million figure, add that definition explicitly. If you can't defend it, replace it with the 18M-to-current growth framing already in the body, and connect that growth to a design mechanism. This is first because it's the only item in the portfolio that risks being read as fabrication rather than imprecision. It affects every archetype and every room this case could reach. Do it before anything else.
2. Add an operating-scope block and connect 3.2× redemption to a decision. Write a single section, modeled on Equinox+'s approach, that separates what you directed, what you delegated, what your three designers owned, and what you escalated. In the same revision, either connect 3.2× redemption to a specific design decision with a mechanism and baseline, or remove it from the hero. If you're short on time, the operating-scope block first. It resolves the IC-director split across every archetype, where the 3.2× fix addresses one metric's credibility. Together they unlock enterprise and player-coach rooms, which currently can't evaluate you from this case because the director title and the IC evidence point in different directions.
3. Add one healthcare-constraint paragraph and fix the 47% label. Document one constraint from the regulated environment and show how it shaped a design decision. In the same pass, reconcile the hero's "47% reactivation" with the body's "47% open rate." The healthcare-constraint paragraph first if you have to choose — it opens a category of roles the case currently can't reach. The 47% label is a five-minute fix you can do in the same session.
The visual-system selection rationale, the missing research trail, the information-model diagram, the structural numbering — all worth doing. But those three revisions are where the current state of the page is actively working against you in rooms you want to be in.
- Gusto's send/review/never-send language: The Head of Design, Unified Service Platform posting asks candidates to define when AI output is ready to ship, needs human review, or should not be sent — the same consequence-literacy gap this audit identifies in Allē's missing healthcare-constraint framing.
- Amplitude's quality-erosion concern: Their Director of Product Design posting explicitly warns that inexpensive AI-generated UI can erode product quality, which makes the visual-system selection rationale (eight directions shown, none evaluated) a more pointed gap if Amplitude is on the target list.
- The CHI progressive-transparency finding: A peer-reviewed study with 12 participants found that eight preferred contextually sufficient transparency over maximal process visibility — relevant to how Allē's "clinically useful" progress record and treatment-timing prompts would need to handle disclosure depth in a regulated context.
- Panel discussion's custody-of-evidence model: The panel proposed an assurance-case structure — claim, argument, evidence, assumptions, boundary, counterevidence — that maps directly onto what this audit finds missing in Allē's outcome-attribution sections.