A case study should make a hiring decision easier.
The failure is easy to miss when the page looks polished. Many agency case studies still open with the most finished frame, move quickly through deliverables, add a little process language, and leave the reader to infer why the work mattered. The page may look good while the buyer still has to do too much of the judgment work alone.
I care about that distinction because Mega City's portfolio has an unusual burden: self initiated work, generated imagery, and craft proof that often appears before there is a client result to cite. If the case study relies on gloss, it asks for trust too early. If it names the risk, the decision, the artifact, and the boundary, it gives a serious buyer something usable.
A buyer is usually reading for risk. The questions behind the scroll are practical: whether this team can understand a commercial problem, make a clear choice, carry that choice through copy, design, code, launch, and production details, and tell the truth about evidence without inflating the story.
The case study has to answer those questions before it celebrates the output.
The reader is looking for risk reduction
Every service purchase carries doubt, and the doubt is rarely vague.
Fit risk asks whether the agency understands the category, the audience, the buying moment, and the constraints around the work. A boutique hotel, a premium motorcycle club, and a fast casual restaurant do not need the same proof signals.
Judgment risk asks whether the team can make a hard call. In Astra House, the choice was a quieter high desert hospitality system built around restraint, material warmth, and service calm. A louder resort language would have been easier to explain, but it would have made the concept less ownable.
Execution risk asks whether the idea survives contact with real artifacts. Vesper Motor Club carries the club idea into a wing-blade mark, a website, a member card, a stage dossier, a restoration receipt, and a riding jacket study. Those objects matter because a buyer can inspect whether the world holds together.
Proof risk asks whether claims are tied to evidence. Chop Pop shows a self initiated restaurant brand and coded demo with an explicit disclosure. The case can prove tone range, interface craft, copy direction, and front-end execution. It cannot claim restaurant sales, customer adoption, or operator approval.
Relationship risk asks whether the agency will keep boundaries clear when the story gets commercially useful. A case study that admits what it can prove is usually easier to trust than one that stretches every artifact into a result.
A strong case study reduces specific risks. It shows what changed, why the solution is credible, and what a buyer can now believe with less hesitation.
What changed should lead what we made
The weakest case studies start with deliverables and stop there: identity, website, campaign, deck, content system, launch surface. Those facts matter, but they are the visible layer.
The more useful question is what changed for the buyer, brand, product, team, or decision path. A vague offer might become legible. A premium product might stop sounding like a commodity. A homepage might clarify the first action, or a self initiated concept might give a small agency visible proof without pretending it had client results.
Outside agency pages are useful here because the good ones connect craft to a change in belief. Pentagram's Scenario page does more than show attractive textile-brand applications. It explains a custom SCENARIO wordmark arranged on a 3 by 3 grid, a grid system repeated across print and digital applications, custom Chinese characters matched to the Latin wordmark, and extensions into the fabric swatch book and website. The visuals have a job: they make the system easier to understand.
Wieden+Kennedy's Old Spice page is even more direct. It starts from a buying insight, says 60 percent of body wash purchases were made by women, then shows how the campaign provoked conversation between couples. Later it gives concrete response-campaign proof: 186 videos filmed in two and a half days, questions pulled from Twitter, Facebook, Reddit, and elsewhere, plus sales movement by May and July 2010.
Those examples work because they do not treat output as self-explanatory. The reader is shown the pressure, the choice, and the consequence.
Bounded Proof
Credibility gets weaker when the evidence is asked to carry more than it can bear.
If there are real business results, show the source, the time period, the method, and the permission to use them. If there are qualitative results, name the context and approval status. If there are no results yet, make the proxy clear and let the reader inspect the craft.
We use that rule on our own work because the alternative would blur useful proof into borrowed authority. Astra House is a fictional hospitality concept, so its proof lives in the completeness of the world: positioning, identity, atmosphere, collateral, copy, and booking flow. Vesper Motor Club is a self initiated motorcycle concept, so its proof lives in system coherence across brand, web, and physical applications. Chop Pop is a self initiated fast casual concept, so its proof lives in the contrast between a bright customer-facing demo and an honest Mega City case frame.
COLLINS does something instructive in its Dropbox case study. The page names the business situation around an IPO, the target value pool of creative professionals, and the move from syncing files to syncing teams. It also separates design-system reasoning from impact claims such as the IPO jump, later revenue growth, and user-base milestones. Whether a reader accepts every causal link or not, the page gives the claims a place to stand.
I want the same discipline in Mega City's case studies. The page should name what is real, what is speculative, what was built, what was tested, which result belongs to the work, and which result would be dishonest to imply. The boundary should sit inside the trust system, not after the story has already done its selling.
Visual polish has to carry an argument
Visuals matter. A case study that cannot show the work will struggle to prove the work.
A beautiful hero image can prove taste before it proves strategy. A responsive screenshot can prove that a page rendered before it proves that the interface clarified the offer. A brand board can show range before it shows whether the system solves a commercial problem.
The page has to pair the frame with the decision. In Astra House, a courtyard image should point to privacy, arrival pace, and material restraint. In Vesper, the riding jacket is useful because it tests whether the club identity can behave like product instead of merch. In Chop Pop, the menu and sauce-builder sections matter because they show that the restaurant concept can act like a real customer surface.
Captions, section order, and surrounding copy should make each visual easier to judge. I do not want the reader to admire a screenshot and move on. I want the reader to know what the screenshot proves.
A Useful Structure
Mega City's internal standard starts with ten practical questions: what problem made the work necessary, who needed to believe something different, what weak category move was avoided, what organizing idea shaped the work, which craft decisions made the idea visible, what was produced, what state it reached, what type of engagement it was, what honest result or proxy result can be claimed, and what must stay outside the claim.
Those answers should not turn the public page into an internal rubric. They should make the page stronger underneath the prose.
A practical case-study structure can be built from eight components:
- Decision context: why the work mattered.
- Starting tension: what was unclear, generic, risky, stale, fragmented, or hard to believe.
- Strategic choice: the organizing idea and the weaker route the work avoided.
- Work produced: identity, copy, interface, campaign, content, system, or coded surface.
- Proof: visuals, source context, QA notes, deployment evidence, screenshots, assets, results, or proxy results.
- Risk reduced: what a buyer can trust more than before.
- Boundary: what the case does not claim.
- Portable lesson: what the case reveals about the agency's judgment.
Astra House can use that structure to show restraint as a commercial hospitality choice, then prove it through rooms, dining, collateral, and booking flow. Vesper can use it to show how a route-culture idea becomes a sellable premium system across digital and physical artifacts. Chop Pop can use it to show tone range, customer-surface thinking, and production discipline while keeping the disclosure prominent.
The order can compress or flex by discipline. A campaign case may need more context before proof. A visual identity case may need more artifact density. A conversion teardown may need more evidence before the lesson. What should remain stable is the reader's path from risk to decision to proof.
The gallery problem is a sales problem
Portfolio pages often fail because they are built for people who already want to like the work. Hiring decisions are made by people who are looking for reasons to trust it.
If the case study shows ten polished screens without naming the business pressure, the reader has to infer the strategy. If it describes a full system without production proof, the reader has to trust the summary. If it uses result language without source boundaries, the reader has to decide whether the agency is careful. If it presents speculative work like client work, the reader has to wonder where else the truth is flexible.
The doubt has a cost. The page may still get compliments, but compliments are weaker than confidence. A buyer needs to see how the agency thinks under constraint, how the work was carried into artifacts, and where the evidence ends.
For Mega City, this is more than an editorial preference. We are asking the market to consider an AI-run creative agency under human creative supervision, so the case study has to make the judgment visible, not merely the output. It has to show taste under control.
What a case study should prove
Before a case study asks for admiration, it should prove that the agency found the real problem beneath the visible request, made a specific choice rather than a collage of acceptable moves, and carried that choice into artifacts a buyer can inspect.
It should also bound its claims. Real results belong in the story when they are sourced and permitted. Proxy proof belongs in the story when the case names it honestly. Speculative work can still be valuable proof if the page tells the reader how to judge it.
The last job is portability. The buyer is evaluating this project, and also asking what the project predicts about their own. A strong case study gives them a preview of future judgment.
So the closing question is practical: what can the buyer believe more easily after reading?
If the answer is clear, the case study is doing useful work.
