Craigslist has worked for decades because it is fast. It is also one of the hardest marketplaces for a newcomer to read. I audited the experience against Nielsen's heuristics, then redesigned one shopping journey: finding a desk, checking the listing and getting to know the seller. The aim was to remove the friction without throwing away the speed.
The homepage lists more than a hundred categories, all at the same visual weight. Someone who has used it for years scans it in seconds. A first-time buyer sees a wall of blue links.
Past the homepage, the problems are smaller but they add up. You open a map and can't find the way out. An icon could mean two different things. A filter label makes you stop and think. Every one of these moments makes the buyer less sure, just when they're deciding whether to trust a stranger.
A clear way in, visual cues about what matters, and an obvious way back when they take a wrong turn.
Density, speed and familiar patterns. Any redesign that slows them down fails the people who already use it most.
How might we make finding, evaluating and contacting sellers about second-hand items more intuitive and trustworthy?
I reviewed the live site using Nielsen Norman Group's ten usability heuristics, rated each issue on the NN/g 0–4 severity scale, and looked at the visual design through three principles.
Do size, weight and layout lead the eye to the most important information first?
Do the things you can click stand out from static text? Craigslist's blue links are the test case.
Are related functions and categories visibly grouped, or just placed next to each other?
Once you're past the homepage, listings are calm and free of distractions. The photo, price and title stand out. The takeaway: refine the typography and spacing, but don't break the simplicity.
Blue underlined links and standard inputs have barely changed in decades. What users have learned never goes out of date. That reliability is worth protecting.
Dense pages let experienced users scan dozens of categories without hidden mega-menus or endless scroll. Any new hierarchy has to keep that speed.
Not everything in the audit made it into the redesign, and not everything in the redesign came from the audit. I've marked which is which.
I prioritised by severity, then by where each issue sits on the desk journey. The map exit was the only Major rating, and it happens right before contact, the moment that matters most for the buyer.
Density itself. The audit showed that much of what makes Craigslist look dated is also what makes it fast. So I didn't remove information. I moved it one step away, and kept text labels and standard controls everywhere.
I mapped the steps before drawing any screens, so each wireframe had one clear job in the journey.
The wireframes tested three ideas early. Could categories live in a dropdown instead of on the page? Could image rows give newcomers an entry point? And could the seller get a page of their own?
On the listing wireframe, "Contact" sits next to the title and repeats under the description, so the main action is always in reach. The contact control opens to a choice of email or phone.
Each screen is shown next to the original, with the finding it answers.
The homepage now starts with real listings and popular searches. The full category list moved into the top navigation, grouped the way buyers already think: Buy, Autos, Housing, Services, Community, Jobs.
Large headings and photo cards give the eye a place to start. The old page offered over a hundred equal options.
The dense list survives inside menus, so experienced users still reach any category in one step.
Navigation uses words, not icons. That's in line with the audit's advice to translate icons into text.
In the original, filters are a long list of checkboxes and empty fields, and it's hard to tell which ones are on. The redesign groups them into labelled sections and repeats every active filter as a chip above the results.
"Within 25 km", "€50–€500", "Good condition". Each can be removed on its own, or all at once with "Clear all".
Location, Type, Price, Condition and Seller each get a heading and a divider. The rarely used options sit under "More filters".
"Search sofas, tables, chairs…" tells you what you can search, and a breadcrumb shows how you got here.
The audit called the original listing page a strength, so I kept its layout: photo first, title and price clear. What changed is how you get around it, a structured set of details, and the seller made visible.
"Home → Furniture → Desks" names every level, so there's no symbol to decode.
"Contact seller" is the only filled button. Save and Share are secondary.
Arrows, thumbnails and a "1 / 6" counter. You always know how many photos are left.
Condition, material, colour and pickup are pulled out of the free-text description.
Name, member since, area and number of active listings. Context before you reach out.
Local pickup advice sits next to the contact action, not on a separate help page.
This fixes the only Major finding. Location is now a dedicated view. A labelled "Back to listing" sits at the top left, with the item's name beside it, and it returns you to the same place on the listing.
"Back to listing" plus the item name. It says where you'll go, not just that you'll leave.
Going back returns you to the same spot on the listing, so the detour costs nothing.
A dashed radius, labelled "not a pickup address", protects the seller and sets honest expectations.
The original has no seller page to compare against, so this part is new. It answers the "trustworthy" half of the design question: who is this, what have others said, and what else do they sell?
Photo, area, member since and a verification badge, with the contact action kept in the same card.
Each review names what was bought ("Purchased: Solid Oak Writing Desk"), so feedback reads as evidence.
A 4-star review mentioning an unlisted scratch is shown alongside the 5-star ones. The breakdown shows the full spread of ratings.
Other items from the same seller turn one purchase into a reason to keep browsing.
This hasn't been tested with users yet, so there are no results to report. Here's what the redesign is meant to improve, and how I'd check it.
The most useful finding wasn't a problem. It was learning which "flaws" users actually depend on.
A heuristic audit can easily turn into a list of everything that looks old. Rating severity forced me to ask who each issue really hurts. That's why the dense homepage got a Minor and not a Major, and why the redesign moves information instead of removing it.
If I did it again, I'd run a short test on the original site before redesigning, so the severity ratings rested on observed behaviour as well as expert judgement.
