Because they are written as project diaries instead of evidence. They describe what the agency did, in order, at length, and never state what changed for the client. A prospect reading it learns about your process and nothing about whether you can solve their problem.
We think the case study page is the most undervalued page on a services website. It is where a warm prospect goes to decide, and it is usually the page with the least care put into it. The homepage gets the budget. The proof gets a screenshot and three paragraphs.
This is how we approach ours, and what the research on trust says about why the order matters more than the writing.
Answer one question: have you solved a problem like mine, for someone like me, with a result I would want? Everything on the page either supports that answer or gets in its way. It is a persuasion page, not a portfolio entry.
The distinction matters because portfolios and case studies get confused constantly. A portfolio shows what things look like. A case study shows what things did. For anyone buying a website to grow a business, the second is the only one that closes work.
The visitor is also further along than most of your traffic. Nobody reads a case study casually. They are checking whether to take you seriously, which means you can be more specific and more direct than you would be anywhere else on the site.
The outcome, stated as a number, above anything else. Not the brief, not the brand, not a wide shot of the homepage. The reader wants to know whether this worked before they invest attention in how it worked.
On our own work page, every project leads with its result. Offbeat Travel shows a 190% rise in organic traffic with a 0.9 second load time. Ajust shows a 230% increase in sign-ups. Sport Design Australia shows a 205% lift in enquiry conversions. Those are the first things you see, before a word of narrative.
We order it that way because it respects how people actually read. A prospect scanning five agencies will give each one a few seconds. If your best fact is in paragraph four, it does not exist.
The context comes immediately after, and it has to be honest. A number without a starting point is decoration. ION Clean Energy went from a 3.4 second load time to 0.8 seconds, and the 3.4 is what makes the 0.8 mean something.
Enough to be credible, not enough to be a project retrospective. A prospect needs the situation, the constraint, the decision you made, and what happened. They do not need your file structure or a week-by-week timeline.
Our rule of thumb is that every paragraph should survive the question "so what?". A sentence about the CMS architecture earns its place if it explains why the client can now publish without a developer. It does not earn its place as a fact about the build.
Specificity is what separates a believable case study from a generic one. "We improved their site speed" is a claim anyone can make. Better Path AI going from 3.1 seconds to 0.7 seconds is checkable, and checkable claims are the only ones that carry weight.
The ones that map to what the client was buying. If they came for leads, show leads. If they came for speed, show speed. Showing an impressive number in a category nobody cares about reads as a dodge, because it usually is.
Where we can, we show both the technical result and the business result, because the two audiences on a case study page want different proof. Dam Secure shows a 185% rise in demo signups alongside a 0.7 second load time. The marketing lead reads the first number. Whoever they have to convince internally reads the second.
Be careful what you promise implicitly. Every number on a case study page is read as an expectation. We only publish results we would be comfortable defending in the first meeting, because a prospect who feels oversold in month one is worse than a prospect you never won.
Say what the number is not, when that matters. A conversion lift measured over three months in a growing market is not the same as a like-for-like comparison, and pretending otherwise is how trust gets spent.
Show the problem and the decision instead. A case study without metrics can still demonstrate judgment, which is much of what a prospect is assessing anyway. What was hard, what you chose, and why, is genuinely useful evidence.
You can also anonymise rather than abandon. Describing the sector, the size, and the shape of the problem without naming the company keeps the substance and respects the confidentiality. Most clients agree to that when they will not agree to a logo.
What we would not do is invent a plausible number or round one up until it sounds better. A case study is a claim about your own honesty as much as your own skill, and it is the worst possible place to be caught stretching.
Because trust is built in stages, and asking for the wrong thing too early breaks it. Nielsen Norman Group's Katie Sherwin described a Hierarchy of Trust in 2016, modelled on Maslow, where users must have basic trust needs met before progressing to more substantial interactions.
Her five levels run from baseline relevance and trust, through interest and preference, then trust with personal information, then trust with sensitive or financial information, and finally an ongoing relationship commitment. Each level depends on the one beneath it.
The rule Sherwin draws from this is the useful part: "Don't make demands at higher levels of commitment until you've addressed all the trust needs at the inferior levels." A case study page that opens with a contact form is asking for level three before it has earned level one.
So the sequence writes itself. Prove relevance, then prove you are better than the alternative, then ask for something. That is exactly why the result goes first and the call to action goes last, and it is the same logic behind good call-to-action design anywhere else on a site.
External confirmation. Nielsen Norman Group's Aurora Harley named being "connected to the rest of the web" as one of four credibility factors in 2016, arguing that businesses and their websites "cannot live in a vacuum" and should link out to third-party sources rather than rely on company-sponsored content.
Applied to case studies, that means a live link to the site you built, a named client rather than an anonymous one where permission allows, and a quote from a real person with a real job title. Each of those is something a sceptical reader can check.
Harley's other factors apply too. Comprehensive and current content matters, which means a case study describing a site that has since been redesigned by someone else needs updating or removing. Stale proof is negative proof.
The client quote is worth getting right rather than getting fast. A specific sentence about what changed in their working week beats a warm sentence about how great you were, and we covered why in our piece on designing social proof people actually believe.
Both, with the individual page doing the real work. A listing page lets a prospect scan for someone like them, which is how most people navigate proof. The individual page is where the decision happens, and it needs to stand alone because that is often the page people are sent.
Individual pages also matter for search, because each one can target the specific combination of industry, platform, and outcome that a prospect would search for. A single listing page cannot rank for ten different problems.
Keep the listing honest about scale. Ten strong case studies presented well beat forty thin ones, and a prospect can tell the difference between a curated set and everything you have ever done.
They are the clearest evidence of first-hand experience on your entire site, which is precisely what search engines and AI systems are trying to assess. A page describing a real project, with real constraints and real outcomes, is not something a competitor can copy or a model can generate about you.
Google's guidance on creating helpful content asks whether it is self-evident who authored the content and whether the material exists primarily to help people. Case studies pass both tests naturally when they are written honestly, because they document something that actually happened.
They also feed the entity picture of your business. Named clients, named sectors, named technologies, and named outcomes tell a model what kind of company you are and who you serve, which is the same job your About page is doing from the other direction.
Move the result to the top of every one. That single change costs an hour and does more than a rewrite, because it puts the deciding fact where the deciding moment happens.
Then check every case study for a live link, a named person, and a date. Anything you cannot support, cut. Anything describing a site that no longer exists in that form, update or retire. A smaller set of current, checkable proof beats a large set of claims nobody can verify.
If you want a second read on whether your proof is doing its job, or you are building a work section from scratch and want it to convert rather than decorate, we are happy to walk through it with you. Let's talk. You can find our team at phoenix.studio.
Tell us where you want to go. We'll tell you how we'd get you there.