Building Case Study Imagery Clients Let You Publish

Ask at kickoff, not at the end

At the start of an engagement everyone is optimistic and the person signing has an incentive to be accommodating. At the end, the champion has moved on, the budget conversation went badly, or the results were mixed and nobody wants their name on it. The same request gets a different answer depending on when you make it.

Put a publication clause in the contract rather than relying on goodwill: you may publish the work in your portfolio and case studies, with the client's name and logo, subject to their review of the specific piece within a defined window. The review window matters more than the permission, because unbounded approval requests die of neglect.

Where the client will not agree to be named, negotiate a fallback in the same conversation: anonymous publication describing the sector and the work without identifying them. Getting the fallback agreed up front is far easier than getting anything agreed later.

Photograph the work, not just the result

Case studies made only of finished deliverables tend to look like a mood board of other people's brands. What makes them convincing is evidence of the work itself: the shoot in progress, the wall of options, the studio setup, the team on location, a screen showing the layout being built. This material is also usually easier to get permission for, because it shows you rather than the client's confidential assets.

Nobody remembers to take these photos in the moment. Add it to the production checklist so someone is responsible for capturing a handful of process shots on every shoot day and at every major milestone. A phone is fine. The value is in having anything at all a year later.

Process imagery has a second benefit: it survives NDAs. A photo of your team setting up lighting reveals nothing confidential about the client and still demonstrates capability.

• Shoot day setup, lighting rigs, the location before it was dressed
• The team working, without confidential material readable on screens
• Selects on a contact sheet, rejected options, the range you produced
• The finished asset in situ: printed, on a shelf, on a phone in a real hand
• Milestone moments like a launch, an install, or a first print run

What to show when the client cannot be named

Anonymised case studies get a bad reputation because they are usually vague. A leading retailer with a well-known problem tells a prospect nothing. The way to make one credible is to be specific about everything except identity: the category, the scale, the constraint, the process, the decisions, and what you produced.

Visually, that means cropping to detail, showing the work at an angle or scale where the brand is not readable, or rebuilding the layout with placeholder branding and saying clearly that it is a reconstruction. What you must not do is present a mockup as if it were a live published asset. A prospect who works out that the screenshot was fabricated will discount everything else on the page.

Where confidentiality is total, some agencies show the process artefacts alone, with a short written account. It reads as honest and it is often more interesting to a sophisticated buyer than another hero image.

Screenshots, mockups and real photography

Each has a role and each has a failure mode. Screenshots are the strongest evidence that something actually shipped, and they age badly as interfaces change. Mockups look polished and prove nothing, and a page of the same three device templates reads as filler. Real photography of the work in the world is the most persuasive and the most effort.

Mix them deliberately rather than by default. One real photograph of the work existing physically does more for credibility than five device mockups. If you use mockups, keep the same template family across a case study so the page holds together, and never let a mockup imply a placement that did not happen.

Screenshots need care at capture. Grab them at high resolution, on a clean profile with no personal data, no unrelated notifications, no test content, and check the small print in the frame before it goes on your site.

Making a year of retainer work look like one story

Retainer work is the hardest to present because it accumulated. Images were shot in different months, by different people, with different cameras, for different channels. Dropped onto one page they look like a folder, not a case study.

Fix it at the presentation layer. Choose a consistent crop set for the page, view the candidates together at final size and cut anything that fights the others, and apply a consistent treatment across the set so they read as a body of work. This is a legitimate use of a per-image editing tool: Foxi AI can bring a mixed set closer to a single presentation, one image at a time, changing lighting and background treatment while the actual work in each frame stays exactly as delivered. What you must not do is edit a deliverable into something you never made.

Then structure the narrative around a small number of moments rather than a chronology. Three well-documented pieces of work with process, decisions and outcome beat twenty thumbnails from twelve months, and they are a lot less work to get approved.

Be careful with results and claims

The imagery is only half the case study; the claims beside it carry the risk. Do not attribute a business outcome to your work unless the client has agreed to that specific claim in writing and the number came from them. Attribution is genuinely hard, clients dislike seeing their numbers published, and a figure you cannot source is the fastest way to lose a sophisticated prospect.

Where you cannot use results, describe what you did and what the client said. A short approved quote from the person who worked with you is often more persuasive than a percentage, and it is far easier to get signed off.

Keep a record of what was approved, by whom and when, alongside the case study assets. People change jobs, and the question of whether you were allowed to publish something comes up years later, usually from someone who was not there.

Frequently asked