Portfolio and case study are different things
A portfolio answers "what have you done." A case study answers "what problem did you solve and what came of it." The reader can't verify the first one: a pretty picture says nothing about how you work or how you'll behave when something goes wrong.
A client choosing a contractor is looking for themselves in your text. They run a salon, a warehouse, or an online school, and they want to see a similar situation - similar business size, similar process mess, similar constraints. Without that, they're left with one comparison criterion: price.
Test your section like this: hide the logos and names. If the cases become indistinguishable, you're not showing work, you're showing wallpaper.
Structure of one case study you can copy
A good case fits into six blocks. You don't need a lot of text - you need order, so the reader can find what they need in half a minute.
The most valuable part here is "what went wrong." A case without a single difficulty looks like an ad. A case that says "the client wanted a site, but it turned out inquiries were getting lost in email, so we first set up forwarding to their CRM" looks like work.
- Who the client is: industry, size, location if relevant.
- What they came with: the brief in the client's words, not "needed to increase conversion."
- What it was before: old site, Excel sheets, inquiries in DMs - how things stood.
- What we built: concrete details, not abstractions. "Built a catalogue of 1,200 items with filters by size and colour, pulled stock levels from their accounting system."
- What went wrong and how we solved it: one honest difficulty.
- What came of it: numbers or clear facts. Link to the live site if it's public.
Numbers: which ones help and which ones hurt
"Conversion grew 300%" without a baseline means nothing - it was two inquiries, now it's six. If you cite percentages, cite absolute values too, plus the period and the data source: analytics platform, CRM panel, client report.
Don't claim all the growth for yourself. The site launched at the same time as an ad campaign - say so. An honest line like "after moving inquiries into the CRM, managers stopped losing requests, according to the client around ten per week" is more convincing than any rounded five percent.
If you have no numbers at all, it's not a death sentence. Non-metric results work too: cut down manual work, the catalogue is now updated by a content manager without a developer, the bot takes inquiries overnight. These are verifiable facts, and they're clearer than unconfirmable statistics.
What to do if the client won't let you talk about it
This is a normal situation: NDA, closed internal tool, or they just don't want the exposure. The project doesn't stop being experience - you can show it anonymously.
Remove the name and logo, replace data with demo content, describe the client by category: "chain of auto repair shops, four locations." You can show the interface with test data or swap real numbers for fictional ones, clearly labelling it as such. One anonymised case with honest mechanics is worth three nameless screenshots.
Permission is easier to get at handoff, while the client is happy with the result. A year later they've moved on to other tasks and an email asking for permission will go unanswered. Ask in writing and clarify boundaries: can you use the name, logo, numbers, screenshots.
Technical mistakes that stop people reading cases
Heavy PNG screenshots at three megabytes each turn the page into a truck. Check your case page in PageSpeed Insights on the mobile tab - if Largest Contentful Paint is over four seconds, some visitors leave before the images appear. Fixed with compression and WebP format.
Second: navigation. If you have more than ten cases, you need filters by task type: e-commerce, landing page, bot, automation. A shop owner shouldn't have to scroll through fifteen landing pages to find one shop.
Third: a dead end at the bottom. Someone who's read the whole case is already warmed up; if there's no button or link to similar work below the text, you lose them at exactly the best moment.
When you don't need case studies
If you do one type of work and sell through personal referrals, a case section might stay an empty formality. Three rough cases are worse than one detailed one - they show you have nothing to say.
Start with one. Take a project you're proud of that had a real difficulty, and describe it using the six blocks above. Then add one case per quarter: a section that gets updated looks alive, and a gallery dumped all at once two years ago does not.
If you have no cases of your own yet at all, honestly write what your experience is made of, and don't fake a portfolio with stock templates. That gets spotted in a minute and costs more than an empty section.
A case convinces not with a picture but with the chain "brief - what was in the way - what we built - what came of it" plus verifiable facts. Start with one detailed case instead of ten screenshots.
Frequently asked
How many cases are enough for a site?
Five or six detailed ones are enough for most studios and contractors. More important to cover different task types: one shop, one landing page, one automation. Identical projects in a row don't add trust.
Can I show projects where the client has since redone the site?
Yes, but say so directly: "site has since been updated by another team, screenshots from 2023." If you give a link without explanation and someone sees a different design, trust evaporates faster than from no link at all.
What do I write in a case if the result wasn't measured?
Describe verifiable facts: what you automated, what the client now does themselves, what process disappeared. Making up percentages is not an option - you'll be asked to back them up on the first call.
Should I do video instead of text?
A short screen recording complements a case well, especially if it's an admin panel or bot. But text gets read and found in search, video doesn't, so it's an addition, not a replacement.