← Issues index

19. Duplicate listing URLs without canonicals — High (PV)

Added 11 Nov 2026. Method: every PV non-article URL with 300+ impressions in Search Console (12 Oct – 8 Nov 2026, 260 URLs) was fetched and grouped by the stories it lists. Files: checks/listing_audit/listing_audit.py, checks/listing_audit/listing_audit_prajavani.json.

What: the same stories are listed at a short URL (/<slug>), a section URL (/<section>/<slug>) and one or more tag URLs. The short URLs have no canonical and carry the site-wide description ("Kannada News - Find Trending Latest News…"). Google picks each as its own canonical and splits searches between them.

Topic URLs listing the same stories (impressions, 28 days) Note
Cartoons (Chinakurali) /cartoons 4.69M (no canonical), /news/cartoons 3.13M, /tags/cartoons 1.88M, /tags/chinakurali 1.41M, /author/prakash-shetty 1.18M, /tags/cartoon 0.76M "chinakurali" (821k impressions) split across 4 URLs; /tags/cartoons 1.88M impressions but 22k clicks.
Churumuri /op-ed/churumuri 3.61M, /churumuri-op-ed 0.76M (no canonical), /tags/ಚುರುಮುರಿ 142k, /tags/churumuri 35k
Horoscope /horoscope 422k and /daily-horoscope 87k (both no canonical), /astro-vastu/horoscope 22k, /tags/horoscope 20k, /tags/horoscope-daily 7k "dina bhavishya" 42.8k impressions at p6.1.
Weekly horoscope /weekly-horoscope 39k (no canonical), /tags/horoscope-weekly 21k, /tags/weekly-horoscope 1k
Panchanga /astro-vastu/panchanga 138k, /panchanga (no canonical)
Districts (~25) /<district> and /<district>-district (no canonical) beside /district/<district>; e.g. /hasana 18.5k p8.8 vs /district/hasana 1.5k; /tumakuru 15.4k p8.9 vs /district/tumakuru p5.7 Short URL usually gets more impressions at a worse position. Also /district-landing, /bengaluru-city-district, /bengaluru-city.
Editorial /abhimata 53k (no canonical, 20 H1s), /editorial 36k (no canonical), /op-ed/editorial 3k
India news /india-news, /india-news-news (no canonical), /news/india-news, /news
Others /chu-bhana 82k vs /columns/old-columns/chu-bhana; /subhashita vs /op-ed/subhashita; /readers-letter vs /op-ed/readers-letter; /short-story vs /art-culture/short-story; /cinema, /movie vs /entertainment/cinema; /articles vs /op-ed/articles; /all-videos, /cinema-photo; /news-landing 466k, /trending 243k (no canonical)

Cause (tpml upstream-sync @ fc684344, 16 Oct 2026; @quintype/seo 1.50.0):

  1. app/server/routes.js:135 sends any /:collectionSlug that isn't a section to PAGE_TYPE.SECTION_PAGE. app/server/data-loaders/section-page-data.js loads the collection with that slug, with no section attached.
  2. In @quintype/seo getSeoData, a section page with no section falls back to getSeoDataFromCollection: canonical = collection metadata["canonical-url"], else SKIP_CANONICAL → no canonical tag. Description = collection summary, else the home page description.
  3. redirectCollectionHandler (app/server/app.js:91), which 301s section-type collections to their section URL, runs only on /collection/:slug, not on /:slug.

How the short URLs come about (checked against PV's public Quintype API, /api/v1/config and /api/v1/collections/<slug>, 11 Nov 2026). The catch-all route (in routes.js since the January 2023 launch build) makes any collection reachable at /<slug>. Three kinds of URL end up there:

Kind Examples Where it comes from Why no canonical Fix
A. A section's own backing collection /hasana-district, /cartoons, /churumuri-op-ed Quintype creates one collection per section (template section); PV's were created on 14 Mar 2023 in the migration. /cartoons is the backing collection of section /news/cartoons. The collection is usually named <section>-<parent section>: 228 of PV's 273 sections have a collection whose name differs from the section's, e.g. /uttara-kannada-district (section /district/uttara-kannada), /india-news-news (/news/india-news), /uttara-dikkininda-old-columns. So most sections have three URLs: /district/uttara-kannada (real), /uttara-kannada (kind B) and /uttara-kannada-district (kind A). The one section rule covers all of them. The loader serves the collection without its section. 301 to the section URL: reuse redirectCollectionHandler for /:slug (it already does exactly this for /collection/:slug).
B. A bare section slug /hasana, /tumakuru, /india-news No collection has that slug, so section-page-data.js finds the section by slug and loads its backing collection, but recurses with sectionObj, which is still {}, instead of the section it just found. The section is found, then dropped. Pass filteredSectionsArray[0] into the recursive call (then the canonical becomes the section URL), or 301 to section-url.
C. Collections created by hand in the CMS /horoscope (created 17 Apr 2023), /daily-horoscope (18 Apr), /news-landing, /abhimata (14 Apr) Made by the team around the April 2023 launch as landing pages (template default, no section). Linked from menus as plain "link" items (menu "collection" items aren't supported by menu-data.js; they render as /#) and from the migration redirect list app/server/handlers/redirect-urls.js (15–19 Apr 2023), e.g. /horoscope.html and /dailyhoroscope.html → /daily-horoscope, /cartoon → /cartoons, /national → /india-news. No section, and no canonical or summary filled in on the collection. Decide one URL per topic; fill in canonical + summary in the CMS, or port DH's generator (self-canonical by default). Point the old redirects at the chosen URL.

Rule order (important). Check the collection first, then the section: (1) a collection with that slug exists and its template is section → kind A, 301 to its section URL; (2) a collection exists with any other template → kind C, leave it; (3) no collection, but a section has that slug → kind B, 301 to the section URL. Checking the section first would be wrong: /horoscope is both a hand-made collection (33,688 clicks in 28 days) and the slug of the section /astro-vastu/horoscope (491 clicks); a section-first rule would redirect the horoscope hub.

Traffic on the URLs that would redirect (Search Console, 12 Oct – 8 Nov 2026, URLs with 300+ impressions; checks/redirect_cleanup/section_rule_impact.json):

Group URLs Clicks Impressions Note
Wave 1: kind B 42 8,029 320,953 Districts, /chu-bhana, /editorial, /india-news… Mostly rank worse than their section URL (e.g. /tumakuru p8.9 vs /district/tumakuru p5.7).
Wave 1: kind A (except cartoons, churumuri) 10 about 1,700 about 150,000 /bengaluru-city-district, /podcast-op-ed, /opinion-op-ed, /india-news-news…
Wave 2: /cartoons (+ ?page=), /churumuri-op-ed 3 2,264,513 5,506,312 Mostly brand searches where the section URL also ranks near 1.
Kind C (not touched by the rule) 13 47,523 1,715,024 /horoscope, /news-landing, /weekly-horoscope, /daily-horoscope, /district-landing, /trending, /abhimata…

Fix:

Fix Where Effort
Fill in canonical URL and summary on each duplicate collection (e.g. horoscope → https://www.prajavani.net/horoscope, hasana → /district/hasana) Quintype CMS, collection metadata; no deploy Small, immediate
Default: when a collection has no canonical, use the page's own URL instead of skipping it customSeo.canonicalUrl in the section loader, or a TPML SEO generator Small code change
301 short URLs to their section (kinds A and B, rule order below) — see Rollout plan app/server/app.js (extend redirectCollectionHandler logic to /:slug) Small; with tests and an on/off switch
Tag pages that duplicate a hub (/tags/cartoons, /tags/horoscope, /tags/chinakurali): canonical to the hub or noindex, follow Tag SEO settings Decide per tag

Rollout options

Option 1: no exceptions Option 2: staged, with exceptions (recommended; detailed below)
What One release: the rule 301s every kind A and B URL, including /cartoons and /churumuri-op-ed. Release 1 301s kind A and B except /cartoons and /churumuri-op-ed; those two get a canonical to their section first, then a 301 once Google has switched.
URLs / clicks moved at once 55 URLs, about 2.27M clicks in 28 days Release 1: 52 URLs, about 10k clicks; the 2.26M move later, after a canonical trial
Time to done 1 release, about 4 weeks to settle 2–3 releases, about 8–10 weeks
Build Simplest: the rule, no exclusion list The rule plus a 2-entry exclusion list and a canonical for those two pages
Main risk If /news/cartoons doesn't take over /cartoons' rankings cleanly, a dip hits the biggest pages at once, and undoing a 301 is slower (Google has to re-learn the old URL). Same risk for the two big pages, but tested first with a canonical, which can be removed with no trace.
Undo Switch the rule off; recovery can take weeks for any URL Google has already swapped. Wave 1: switch off. Wave 2 canonical: remove it.
Same in both Rule order (collection first, then section); kind C untouched; query string kept; #25 cleanup in the same release; baseline before release; tests; on/off switch.

Option 2 in detail (the rollout plan agreed 11 Nov 2026):

Step What When Undo
0. Baseline Record clicks, impressions and position for every URL that will change and its section URL, plus the brand searches "prajavani cartoon", "chinakurali", "churumuri" (the weekly scorecard in Content Engine/build-plan.md, item 0). Before any release —
1. Build One rule on /:slug in the order above (collection first, then section), behind an on/off switch. Keep the query string (/cartoons?page=9 → /news/cartoons?page=9), compare slugs in lower case, and exclude /cartoons and /churumuri-op-ed (wave 2). Ship the #25 redirect cleanup in the same release so no hand-written redirect chains through the rule. Release 1 Switch off
2. Test Before release: run the rule against a list of URLs: every kind A and B URL → its section URL in one 301; every kind C URL (/horoscope, /news-landing, /daily-horoscope…) and top-level section untouched; no loops; check_redirects.py all one-hop. Release 1 —
3. Wave 1 Kind B (42 URLs, 8,029 clicks in 28 days) and kind A except cartoons and churumuri (10 URLs, about 1,700 clicks): straight 301 (no canonical step, no 302). Release 1 Switch off
4. Watch 3–4 weeks: section URL clicks vs old URL clicks combined; URL Inspection on a sample of redirected URLs; a dip of 1–3 weeks is normal. Weeks 1–4 —
5. Wave 2, canonical first /cartoons and /churumuri-op-ed (2.26M clicks): set their canonical to /news/cartoons and /op-ed/churumuri (CMS collection canonical or code), not a redirect yet. Watch 2–3 weeks until Google shows the section URLs for the brand searches. After wave 1 is stable Remove the canonical
6. Wave 2, 301 Remove the exclusion so the rule 301s them too. When step 5 shows the section URLs ranking Restore the exclusion
— Kind C (/horoscope, /news-landing, /daily-horoscope, /district-landing, /trending, /abhimata…; 47,523 clicks) Separate editorial decision —

Risks (a 301 is the standard, safe way to consolidate duplicates, but not zero risk):

  1. Re-processing dip: 1–4 weeks of fluctuation while Google swaps URLs. Expected; should recover.
  2. Consolidated URL ranks lower than the old one did: the main lasting risk, and mostly for /cartoons, which Google currently prefers (2.19M clicks vs 459k for /news/cartoons). Mitigation: canonical-first for wave 2 (step 5), which is reversible with no trace.
  3. Different title/description on the section URL can change the share of people who click at the same position.
  4. Implementation bugs (wrong pages redirected, as /horoscope nearly would have been with a section-first rule; dropped query strings; loops): the most likely way to lose traffic. Mitigation: steps 1–2 and the switch.
  5. Non-search traffic (Discover, apps, partners, bookmarks) still lands via the redirect, but anything that reports by URL will show the old URL dropping.

Lower-risk alternative for wave 2: leave /cartoons and /churumuri-op-ed as they are, with a canonical pointing to themselves (CMS or code). That ends the "no canonical" state with no risk to their 2.26M clicks, but keeps the duplicate with /news/cartoons and /op-ed/churumuri. Wave 1 alone fixes most of the URLs.

Verify: view-source shows one canonical on every short URL; URL Inspection shows Google's canonical as the chosen hub; split queries in Search Console converge on one URL.

DH: not affected, and how DH fixed it. The same audit on DH (228 non-article URLs with 300+ impressions; checks/listing_audit/listing_audit_deccanherald.json) found only 2 without a canonical (/news_sitemap.xml, /grievance-redressal). DH has the same /:collectionSlug catch-all route, and its locked @quintype/seo (1.42.2) is worse than PV's: its collection fallback always skips the canonical, even if one is set in the CMS. DH avoids this with its own SEO generator for section pages, app/isomorphic/components/utils/custom-collection-tags.js (wired in app/server/app.js, customizedTags), which always outputs a canonical: the section URL, else sketches-host + the collection slug.

Repos: GitHub access for product-TPML returned 403/404 on 11 Nov 2026, so neither repo could be synced; analysis used the last fetched mains (PV fc684344, 16 Oct; DH 0bc245a1, 23 Oct 2026).

Update 19 Nov 2026: full URL inventory, both sites (checks/listing_audit/short_url_inventory.py; outputs short_url_inventory_prajavani_2026-11-19.csv/.md and short_url_inventory_deccanherald_2026-11-19.csv/.md). Every one-segment URL the /:slug route can serve, from the code's own rules (Quintype config: 273 PV / 174 DH sections and their backing collections) plus every one-segment URL in Search Console (web + Discover, 20 Jul 2025 – 16 Nov 2026), the menus and the Page indexing exports. Slugs Google or the menus know were classified with the collections API (collection first, as the router does; origin host, not Cloudflare). Live spot check of 6 random URLs per site (Googlebot UA).

Site Kind Action URLs Known to Google Clicks (16 months) Impressions
PV A: section's backing collection 301 to section URL 238 93 48,437,192 147,549,420
PV of which /cartoons + /churumuri-op-ed (wave 2) canonical trial, then 301 2 2 48,352,845 141,883,892
PV B: bare section slug 301 to section URL 212 154 214,421 6,181,687
PV C: hand-made collection keep; own canonical and description 50 50 943,536 41,763,680
DH A canonical (or 301) to section URL 140 41 31,760 2,164,316
DH B canonical (or 301) to section URL 100 8 6,280 557,816
DH C keep (already self-canonical) 102 102 1,562,840 53,667,631