A stranger in a hurry
/tag pagesOne page, four moves
1 · Build authored folklore pages
A few genuinely authored, uniquely illustrated pages on proven-demand folk characters — not programmatic tag pages. The craft is the moat.
2 · Pilot with Штирлиц
Best learning per unit of risk: existing tag baseline to measure against, green copyright, art already on the site.
3 · Returns off-site
The audience is Telegram-native and never logs in. Drive return via a Telegram channel + a daily hook, not on-site retention gimmicks.
4 · Measure honestly
Beat the site's own baseline: below 62.5% bounce, above 2.0 jokes/session, nonzero internal click-through.
Nine studies →
Who actually visits
Measured from live Яндекс.Метрика (counter 108508005) + the production database. Return and depth are the weakest points — and what these pages must move.
The core visitor
A first-time, mobile, Yandex-arriving Russian adult (skews male ~30–55), Russophone incl. a CIS/diaspora tail, who wants a laugh now with zero friction — no signup, no reading assignment.
What buys 3–10 more minutes
A page that promises and delivers "the best of <something they recognize>" fast, with light atmosphere — not a biography wall. Triggers: the definitive-best, recognition, nostalgia (40+ only), shareability.
Demand and copyright point at the same targets
The characters people search for are anonymous folklore — free under Russian law. The famous humorists carry both weaker joke-demand and live copyright. Safest content = most wanted.
What we can legally use
A practical risk framework grounded in the primary text of the Russian Civil Code, Часть IV. Not legal advice — per-item clearance still required; route red uses to the operator.
Green freely
- Anonymous folk anekdots (ст. 1259 п.6)
- Facts & history in our words
- Commissioned original art
- Ильф и Петров — PD since 2017
Yellow care
- Short attributed quotes of a protected author (ст. 1274)
- Uncertain provenance → treat as authored
- Character whose source work is protected
Red avoid
- An author's material as the page's substance
- Soviet photos, film stills, A/V (+URAA)
- A page that's really a collection
Safest sources to start
- lib.ru /ANEKDOTY/ — canonical folk anekdots («1001 сов. полит.»)
- ru.Wikisource — public-domain text, built for reuse
- Academic books (Graham, Shturman-Tiktin, Borev) — certify folklore + citable framing
- Prokudin-Gorsky @ LoC — the one clean PD image well
The media trap
Assume every Soviet photo, film still, poster, and clip is owned until proven otherwise (2008 retroactivity + URAA for US hosting). The safe path is what the site already does for comics: commission original art, and embed official A/V rather than host it.
Steal the mechanic, not the look
None of the five is a comedy site — that's the point. Borrow the interaction principle, reinterpret through Anekdotnik + Soviet context + the subject's personality.
Project 1917
The legal-and-structural model: real figures via short attributed excerpts, never whole works. Each card standalone & shareable.
Internet Artifacts
One joke-sized card per scroll-stop on an era rail. The cheapest, closest page skeleton — mobile-native.
The Boat
Atmosphere-per-kilobyte: hand-drawn panels, scroll-paced reveals, one mute-able sound. The warm/not-kitsch target.
Gucci Garden tour
The "enter a Soviet room" mechanic — a radio, a TV, a newspaper as tappable objects — without heavy 360° panoramas.
Google Arts "Story"
The Instagram-Stories pattern the audience already knows: one image + one line per slide, into the joke feed.
Avoid the 2026 3D trend
WebGL showpieces win awards but exclude the low-end Android phones this audience holds. Borrow the taste & restraint of luxury sites, never their engines.
What to build first
A weighted 14-criteria score — copyright safety (×2) and content availability (×1.5) are gating. Top five in orange; recommended pilot ★.
Штирлиц, as an absurd serialized show
Not a ranked list — the ~618 folk jokes curated into mini-episodes by recurring motif, each joke still a standalone card.
Why this pilot (not the famous name)
- Existing
/tagbaseline → a clean before/after test - Green copyright + banner art already on the site
- High design potential → actually tests "does atmosphere beat a thin list?"
- Enough demand (7.1k/mo) for signal at low traffic
Prod-safe delivery
- Standalone route, isolated — no change to feed, tags, ranker, logic
- Read-only over joke data; chapters are a manifest, not new tags
- Ships as a Vercel preview link, not on prod
- No film stills — folklore + commissioned art only
Mechanics for THIS audience
Classified for a first-time, login-less, mobile audience — not because modern sites use them.
High impact
- Frictionless on-theme next-joke loop
- Same-decade random · next character · related chains
- Sharing & voting (already built)
Experimental
- Furniture-as-navigation (only if it never slows browsing)
- Daily joke · "remember this one?" · easter eggs
Skip / harmful
- Like-based personalization — measurably unbuildable at 98.5% first-time
- Comments (moderation cost, no community)
- Streaks / login-nagging (dark pattern)
Return is mostly an off-site job
With no login and a ~1.5% return baseline, on-site retention features assume loyalty that isn't there yet. Meet the audience where its habit already is.
1 · Telegram channel
The audience is Telegram-native and the site already shares there. Turn "return to the website" into "see it in a feed you already scroll." The single biggest return bet.
2 · A daily-fresh hook
«анекдот дня» / «в этот день» — a reason the page is different tomorrow, and the channel's daily payload.
3 · Shareable objects
A quote card, a character card worth screenshotting. For a login-less audience, return is often someone else's share.
Maintainable by a small team
One automated daily rotation + a planned editorial/seasonal drop calendar — not per-user retention machinery.
How it plugs into the real site
A thematic page is mostly a query over the existing ~43k jokes, not new scraping. It reuses the joke card, votes, share, tracking and taxonomy — strengthening the central DB, never a silo.
Ingestion pipeline (reuses existing tooling + two gates)
The dominant job is re-tagging jokes we already have with character/era metadata; external top-up (lib.ru folklore) is light and human-reviewed. Automatic ingestion only for confirmed folklore, and it still lands as pending.
Discovery graph — half built already
The taxonomy already carries facet and a related[] edge. Add era + person facets and it scales as tag queries over the existing GIN index. Links via a small human-worded "discovery rail," never a link farm.
SEO thesis
The site's own data: thin topic pages draw ~0 and bounce 62.5%. So editorial uniqueness is the only thing that makes a topic page rank — a few genuinely unique pages, canonicalized against legacy /tag, richly cross-linked. Not hundreds of programmatic pages.
Analytics — success against the site's own baseline
| KPI | Baseline (measured) | Target |
|---|---|---|
| Jokes consumed / session | ~1.3 | ≥ 2.0 |
| Bounce / quick-exit | 62.5% (tag pages) | < 40% |
| Internal click-through (theme→joke/theme) | ~0 | > 0, rising |
| Next-day / 7-day return | ~1.5% | directional (thin traffic) |
First test = an upgrade-in-place holdout: the thematic page vs the existing thin tag page on the same intent. Return-rate A/Bs are underpowered at ~57 organic visits/day — stated honestly.