Article indexes people actually browse

Open ten agency blogs and nine of them are the same page: a hero card for the newest post, then an infinite column of identical cards marching back to 2019, sorted by date and nothing else. It is a conveyor belt. It tells the reader one thing — "things happen here, sometimes" — and it answers none of the questions a real visitor arrives with. Is this place for me? Where's the good stuff? How deep does this go?

An article index is not a database dump with card chrome. It's the front door of your thinking. We design it the way a good magazine editor lays out a contents page: with opinion, rhythm, and a clear sense of what deserves your attention first. Here's the playbook we use, including the decisions on this site's own journal.

The conveyor belt problem

Reverse chronological order is a fine default for a database and a lazy one for a reader. Date order says "newest is best", which is rarely true — your best piece might be eighteen months old, and your newest might be a minor release note. It also flattens everything to one visual weight, so a 2,000-word flagship essay sits next to a four-paragraph announcement wearing the same card.

The failure mode we see most in audits: indexes that answer "what did you publish lately?" when visitors are asking "what do you know about X?" A prospective client landing on your journal wants to trace a path — from a topic they care about, to evidence you've thought about it hard, to the work that proves it. Date order makes them do that archaeology themselves. Most won't.

The fix is not abandoning chronology — recency is a real signal that the studio is alive. The fix is layering editorial structure on top of it.

Lead with a lead story

Newspapers worked this out a century ago: one story gets the big treatment, and everything else falls in behind. On a journal index, the lead story card does three jobs at once. It tells a first-time visitor what you consider your best current thinking. It gives returning readers a reason to look (the lead rotates). And it breaks the monotony of the card grid, giving the page a beginning.

Our rules for the lead slot:

One lead per page-one view, set significantly larger than the grid cards — display-weight title, a real dek, an image or strong typographic tile. On our journal it's the latest article that has a proper hero image, which quietly enforces a quality bar: if a piece didn't earn a hero, it doesn't earn the lead.
Exclude it from the grid. Nothing says "template" like the lead story appearing twice on the same page — once enormous, once as card number one. Dedupe at build time, not in the stylesheet.
Page one only. Pagination pages two and beyond drop straight into the grid. The reader who clicked "page 3" is browsing the archive; ceremony is finished.

Rhythm: card, list, card

A wall of identical cards reads as wallpaper after the second row. The eye needs syncopation. On dense indexes we mix two or three densities: full cards with images for pieces that have them, slim typographic rows (title, meta, arrow) for shorter reads, and the occasional full-width feature. Our resources hub, for instance, uses compact numbered rows — a list that behaves like a table of contents rather than a Pinterest board.

A practical test: blur your eyes and look at the page. If you see one texture from top to bottom, you have wallpaper. A good index has a readable rhythm at blurry-vision distance — heavy, light, light, medium. This is the same instinct behind good editorial grid work: rhythm first, rules second.

Metadata: orient, don't fill

Every card carries a metadata decision, and the temptation is to carry everything: author, date, reading time, tags, category, a share icon nobody has ever clicked. Resist. Card metadata exists to help a reader choose, and the signals that genuinely help are few:

Cluster or category — "Engineering" tells a skimming founder whether this card is her conversation. One label, set quietly.
Reading time — a promise a tired reader can act on. "6 min" converts better than any teaser copy we've tested against it.
Date — matters disproportionately for fast-moving topics (AI, platform changes) and barely at all for craft pieces. We show it; we don't lead with it.

Descriptions deserve a harder look. A dek written for humans — a genuine one-line argument for reading — lifts click-through measurably. A dek that's the first sentence of the article with its knees cut off subtracts trust. Our rule: if the writer didn't write the description, the card doesn't show one. This holds down on case study listings too, where the client's industry and services orient faster than any excerpt.

Beyond page one: pagination without fatigue

Somewhere past sixty or eighty entries, every index faces the scale question. The usual answers are infinite scroll (a trap — it kills the footer, wrecks deep-linking, and makes "I saw it about halfway down" unrecoverable) or endless pages of twelve. Both treat the archive as a queue. Readers experience it as a swamp.

What works at scale:

Filters that earn their place. Cluster or tag filters let a reader say "only show me the growth pieces" — but only if the taxonomy is honest and every option has a count. A filter that yields zero results is a broken promise in a pill. Show counts, URL-sync the state (the URL is your best state manager — our engineers wrote a whole piece on why), so a filtered view is shareable and back-button-safe.
Route-based pagination, not a load-more button. Real pages with real URLs mean page 4 can be linked, indexed, and returned to. It also forces a discipline: if page 9 of your journal is embarrassing, the problem is your archive, not your pagination.
Internal navigation between pages that respects the reader's position. "Older / Newer" with page numbers, not a carousel of "1 2 3 … 47". Nobody has ever clicked 31.

If your index has hundreds of entries, also give people an exit ramp that isn't the index: a strong search with an empty state that shows popular tags and latest articles. Browse and search are different moods; serve both.

The index is a argument for the studio

This is the part agencies forget while styling cards. Your article index is often the second page a prospective client visits, and it makes an argument whether you intend it to or not. A blog with three posts from 2023 argues you stopped thinking. A wall of uncategorised listicles argues you think in volume. An index with a confident lead story, honest filters, and visible depth argues: these people have a practice, and it's been running for years.

That argument is convertable. Route it somewhere. The end of a good cluster page should gesture at the adjacent commercial surface — reading about type systems should end one click from how we build websites. Not a hard sell; a door left open. We place related-case-study links inside the index furniture rather than a shouting CTA banner, because the reader came to browse, and manners matter. (Our piece on related-content modules covers the ranking rules in detail.)

A pre-flight checklist for index pages

Before we ship any article index, we run it through this list:

Does page one have a clear lead story, excluded from the grid?
Does the page have rhythm at blurry-vision distance?
Is every metadata element on every card a genuine choosing signal?
Do all filters show counts, URL-sync, and never dead-end at zero?
Can page 4 be linked, shared, and crawled?
Does the archive tell the truth about volume — no fake "load more" that ends after two clicks?
Does the index route browsing readers one respectful step toward services or work?

Key takeaways

Reverse chronological is a storage order, not a reading order. Layer editorial structure over it.
One lead story, big, deduped from the grid, page one only.
Mix densities — cards, rows, features — so the page has rhythm instead of wallpaper.
Card metadata exists to help readers choose: cluster, reading time, date. Descriptions only if a human wrote them.
Filters need counts and URL state; pagination needs real routes; search needs a good empty state.
Your index makes an argument about your practice. Make it on purpose, and leave a door open to the commercial pages.

Frequently asked questions

Should the newest article always be the lead story?

No — the newest article with a proper hero and a full argument should be. Tying the lead slot to an asset (a hero image, a written description) is a forcing function for quality. A thin announcement as your lead story tells visitors your best thinking is an announcement.

Isn't infinite scroll better for engagement?

It inflates scroll-depth metrics and quietly destroys everything else: footers, deep links, back-button behaviour, and the reader's sense of the archive's size. On content sites we've audited, infinite scroll raised "pages per session" while lowering continuation to actual articles. Use real pages.

How many cards per page?

We default to twelve plus the lead. Enough to feel deep, few enough that page two isn't a junk drawer, and divisible by two and three so the grid never orphans a final card. The number matters less than the rule: pick one, hold it, and let the pagination routes fall out naturally.

Do tags or clusters work better as the primary filter?

Clusters as primary (five to nine of them, matching your actual practice areas), tags as the serendipity layer. Tags multiply faster than any governance can control; clusters make a promise about who you are. If a reader can't guess what's in a cluster from its name, rename it before you ship the filter.

Our archive is thin — should we hide the index until it's bigger?

No, but be honest about scale. Twelve good pieces with a confident lead story and no fake pagination beats forty pieces with bulk filler. A small, sharp archive argues focus. A padded one argues the opposite.